STATS SA 03-10-01... · Findings from the Living Conditions Surveys, 2008/2009 - 2014/2015 Report 03-10-01. Koch Street, Salvokop, Pretoria, 0002 ... is not sufficient to capture

STATISTICS SOUTH AFRICA

STATS SA

THE SOUTH AFRICA I KNOW, THE HOME I UNDERSTAND

Subjective Poverty in South Africa

Findings from the Living Conditions Surveys, 2008/2009 - 2014/2015

Report 03-10-01

Koch Street, Salvokop, Pretoria, 0002 Private Bag X44, Pretoria, 0001, South Africa User information service: +27(12) 310 8600, Fax: +27(12) 310 8944 Main switchboard: +27(12) 310 8911, Fax: +27(12) 310 8944 Website: www.statssa.gov.za, Email: [email protected]

Subjective Poverty in South Africa Findings from the Living Conditions Survey 2008/2009 – 2014/2015

Statistics South Africa, 2018 Risenga Maluleke, Statistician-General

Report No.03-10-01

Subjective Poverty in South Africa: Findings of the Living Conditions Survey, 2008/2009 – 2014/2015 / Statistics South Africa Published by Statistics South Africa, Private Bag X44, Pretoria 0001 © Statistics South Africa, 2018 Users may apply or process this data, provided Statistics South Africa (Stats SA) is acknowledged as the original source of the data; that it is specified that the application and/or analysis is the result of the user's independent processing of the data; and that neither the basic data nor any reprocessed version or application thereof may be sold or offered for sale in any form whatsoever without prior permission from Stats SA. Stats SA Library Cataloguing-in-Publication (CIP) Data Subjective Poverty in South Africa: Findings of the Living Conditions Survey, 2008/2009 – 2014/2015 / Statistics South Africa. Pretoria: Statistics South Africa, 2018 Report No. 03-10-01 94 pp ISBN 978-0-621-46261-6 A complete set of Stats SA publications is available at Stats SA Library and the following libraries:

National Library of South Africa, Pretoria Division National Library of South Africa, Cape Town Division Library of Parliament, Cape Town Bloemfontein Public Library Natal Society Library, Pietermaritzburg Johannesburg Public Library Eastern Cape Library Services, King William’s Town Central Regional Library, Polokwane Central Reference Library, Mbombela Central Reference Collection, Kimberley Central Reference Library, Mmabatho

This report is available on the Stats SA website: www.statssa.gov.za For technical enquiries please contact: Rodney Khumalo Tel: 012 337 6394 Email: [email protected]

Table of contents

Chapter 1: Introduction and background 2 1.1 Introduction 3 1.2 Measures of poverty 3 1.3 Objectives of the report 6 1.4 Methodology and data sources 6

1.4.1 Data sources 6 1.4.2 Methods 7

1.5 Limitations of the report 8 1.6 Layout of the report 9

Chapter 2: Key findings 12 2.1 Decline in subjective poverty in South Africa between 2009 and 2015 13 2.2 A comparison of subjective poverty levels and objective income poverty levels in 2009 and 2015 14 2.3 Similarities between subjective poverty and objective poverty in 2009 and 2015 15 2.4 Socially perceived necessities and subjective poverty status 16

Chapter 3: Self-perceived wealth indicator 20 3.1 Profiling subjective poverty using the self-perceived wealth question in South Africa; 2009 and 2015 21 3.2 Comparing subjective poverty levels (SPWQ) and objective income poverty levels; 2015 25 3.3 Subjective poverty profile using Socially Perceived Wealth Indicator (SPWQ) 29

3.3.1 Individuals identified as subjectively poor but measured as objectively non-poor. 29 3.3.2 Individuals classified as subjectively non-poor but measured as objectively poor 33

3.4 Determinants of self-perceived wealth indicator 35

Chapter 4: Minimum income indicator 38 4.1 Profile of subjective poverty (MIQ) in South Africa; 2009 and 2015 39 4.2 Comparing subjective poverty levels (MIQ) and objective income poverty levels; 2015 42 4.3 Determinants of minimum income indicator 51

Chapter 5: Income evaluation indicator 54 5.1 Profile of subjective poverty (IEQ) in South Africa; 2009 and 2015 55 5.2 Comparing subjective poverty levels (IEQ) and objective income poverty levels; 2015 58 5.3 Determinants of income evaluation indicator 63

Chapter 6: Living circumstances of households 66 6.1 Household expenditure 67 6.2 Living circumstances of households 70

Chapter 7: Summary and conclusion 76

Annexure A 84

List of tables

Table 1.1: Inflation-adjusted poverty lines, 2009 to 2015 (per person per month in Rand) 4 Table 1.2: Subjective poverty indicators available in the LCS 2008/09 and LCS 2014/15 8 Table 2.1: Proportion of individuals and households who perceived themselves as poor in 2009 and 2015 13 Table 2.2: Subjective and objective poverty headcount rates; 2009 and 2015 15 Table 2.3: Distribution of individuals by poverty status based on objective and subjective poverty

measures; 2009 and 2015 15 Table 2.4: Proportion of individuals in households with access to socially perceived necessities by

subjective poverty status; 2015 16 Table 2.5: Proportion of individuals in neighbourhood with access to socially perceived necessities by

subjective poverty status; 2015 17 Table 3.1: Proportion of subjectively poor (SPWQ) by population group; 2009 and 2015 22 Table 3.2: Proportion of subjectively poor (SPWQ) by province; 2009 and 2015 23 Table 3.3: Proportion of subjectively poor (SPWQ) by type of settlement; 2009 and 2015 24 Table 3.4: Proportion of subjectively poor (SPWQ) by age and sex of head of household; 2015 24 Table 3.5: Percentage distribution of households by poverty status based on the LBPL line and the

subjective poverty measure; 2015 26 Table 3.6: Percentage share of poverty by province and poverty indicator; 2015 26 Table 3.7: Percentage share of poverty by settlement type and poverty indicator; 2015 27 Table 3.8: Percentage share of poverty by population group and poverty indicator; 2015 28 Table 3.9: Poverty headcount rates by highest education level of household head; 2009 and 2015 29 Table 3.10: Percentage distribution of individuals and households classified as subjectively poor but

measured as objectively non-poor; 2009 and 2015 30 Table 3.11: Percentage share of individuals classified as subjectively poor but measured as objectively non-

poor, by socially perceived necessity of regular saving for emergencies; 2009 and 2015 30 Table 3.12: Percentage share of individuals classified as subjectively poor but measured as objectively non-

poor, by socially perceived necessity of having a bath or shower in the house; 2009 and 2015 31 Table 3.13: Percentage share of individuals classified as subjectively poor but measured as objectively non-

poor, by socially perceived necessity of having somewhere for children to play safely outside the house; 2009 and 2015 32

Table 3.14: Percentage distribution of individuals and households classified as subjectively non-poor but measured as objectively poor; 2009 and 2015 33

Table 3.15: Percentage share of individuals classified as subjectively non-poor but measured as objectively poor, by socially perceived necessity of regular saving for emergencies; 2009 and 2015 33

Table 3.16: Percentage share of individuals classified as subjectively non-poor but measured as objectively poor, by socially perceived necessity of having a bath or shower in the house; 2009 and 2015 34

Table 3.17: Percentage share of individuals classified as subjectively non-poor but measured as objectively poor, by socially perceived necessity of having somewhere for children to play safely outside the house; 2009 and 2015 34

Table 4.1: Proportion of subjectively poor (MIQ) by population group; 2009 and 2015 40 Table 4.2: Proportion of subjectively poor (MIQ) by province; 2009 and 2015 40 Table 4.3: Proportion of subjectively poor (MIQ) by type of settlement; 2009 and 2015 41 Table 4.4: Proportion of subjectively poor (MIQ) by age and sex of head of household; 2015 42 Table 4.5: Poverty indicators by province; 2015 44 Table 4.6: Poverty estimates by settlement type; 2015 45 Table 4.7: Poverty estimates by sex of the household head; 2015 46 Table 4.8: Poverty estimates by population group; 2015 47 Table 4.9: Distribution of households by poverty status based on the lower-bound poverty line and

subjective poverty measure; 2015 47 Table 4.10: Percentage share of poverty by province and MIQ poverty indicator; 2015 48 Table 4.11: Percentage share of poverty by settlement type and MIQ poverty indicator; 2015 49 Table 4.12: Percentage share of poverty by population group and poverty indicator; 2015 50 Table 5.1: Proportion of subjectively poor (IEQ) by population group; 2009 and 2015 55 Table 5.2: Proportions of subjectively poor (IEQ) by province; 2009 and 2015 56 Table 5.3: Proportion of subjectively poor (IEQ) by type of settlement; 2009 and 2015 57 Table 5.4: Proportion of subjectively poor (IEQ) by age and sex of head of household; 2015 58 Table 5.5: Distribution of households by poverty status based on the lower-bound poverty line and the

subjective poverty measure; 2015 59 Table 5.6: Percentage share of poverty by province and poverty indicator; 2015 60 Table 5.7: Percentage share of poverty by settlement type and poverty indicator; 2015 61 Table 5.8: Percentage share of poverty by population group and poverty indicator; 2015 62 Table 6.1: Average annual household expenditure among the poor by poverty indicator (in Rand); 2015 69 Table: A.1 Logit regressions of South Africans by poverty status and subjective poverty indicator; 2015 85

List of figures

Figure 2.1: Proportion of subjective poverty by population group; 2009 and 2015 14 Figure 3.1: Poverty headcount by poverty line used; 2015 25 Figure 4.1: Poverty headcount by poverty line used; 2015 43 Figure 5.1: Poverty headcount rates by poverty line used; 2015 59 Figure 6.1: Proportion of households that received a government housing subsidy or RDP house by

province and poverty indicator; 2015 70 Figure 6.2: Proportion of households who have access to piped water inside their dwellings or on-site by

province and poverty indicator; 2015 71 Figure 6.3: Proportion of households that had access to flush toilets inside their dwellings or on-site by

province and poverty indicator; 2015 72 Figure 6.4: Proportion of households living within a 2km radius of selected facilities by type of facility and

poverty indicator; 2015 73 Figure 6.5: Percentage of households in which a member of the household ‘never goes hungry’ by

population group and poverty indicator; 2015 74

Preface

This report presents the subjective poverty levels based on data collected by Stats SA through the Living Conditions Survey (LCS). The first report on subjective poverty in South Africa by Stats SA was released in 2012 with assistance from the School of Development Studies at the University of KwaZulu-Natal using the data collected in 2008/09. This current report builds on the first report utilising the rebased national poverty lines published in 2015. The main purpose of this current report is to provide an updated subjective poverty profile of South Africa using the most recent LCS data collected in 2014/15. Risenga Maluleke Statistician-General

Chapter 1Introduction and background

Chapter 1Introduction and background

3

1.1 Introduction

Poverty is a complex concept, which may include social, economic, and political elements. Poverty is usually defined as the deprivation of the means necessary to meet basic needs such as food, shelter and clothing (Jansen et al, 2013). Income has been critical in the measure of poverty. Poverty has been commonly measured in absolute or relative monetary terms. However, there has been a lot of debate about the best approach to measure poverty because the poor themselves consider their experience of poverty much more broadly. Therefore focusing on one factor alone such as income, is not sufficient to capture the true reality of poverty. Statistics South Africa (Stats SA), over the years, has produced a number of reports based on various definitions and measures of poverty. These include objective measures, relative measures, multidimensional measures as well as subjective poverty measures. Subjective poverty as the main focus of this report is defined as the self-assessed economic status of people relative to others. It can also be perceived consumption adequacy, where people can reflect on the minimum income they need to make ends meet (Posel and Rogan, 2014). This report, firstly provides an overview of different measures of poverty that Statistics South Africa have used in the past, which amongst others, include: objective poverty, relative poverty and multidimensional poverty measures; as a preamble to the in-depth discussing of subjective measures of poverty.

1.2 Measures of poverty

Objective poverty measures are based on an expert derived definition of poverty, where a poverty line (either based on income or expenditure) is derived and people are classified as poor if their reported income/expenditure falls below the poverty line; or non-poor if it falls above the poverty line. This method is best used for monitoring of progress and reporting for policy purposes. The line is standard, so progress (or non-progress for that matter) can easily be determined. Stats SA published the country’s first official national poverty lines in 2012 where the cost-of-basic needs approach was used to produce three poverty lines namely: the food poverty line, the lower-bound poverty line and the upper-bound poverty line. These lines capture different degrees of poverty and allow the country to measure and monitor poverty at different levels. The food poverty line is the Rand value below which individuals are unable to purchase or consume enough food to supply them with the minimum-per-capita-per-day energy requirement for adequate health. The lower-bound and upper-bound poverty lines are derived using the food poverty lines as a base, but also include a non-food component. Individuals at the lower-bound poverty lines do not have command over enough resources to enable them to purchase or consume both adequate food and non-food items and are therefore forced to

4

sacrifice food to obtain essential non-food items. Meanwhile, individuals at the upper-bound poverty line can purchase both adequate levels of food and non-food items. Table 1.1: Inflation-adjusted poverty lines, 2009 to 2015 (per person per month in Rand)

Year* Food-poverty line (FPL) Lower-bound poverty line (LBPL) Upper-bound poverty line (UBPL)

2009 318 456 709

2010 320 466 733

2011 335 501 779

2012 366 541 834

2013 386 572 883

2014 417 613 942

2015 (April) 441 647 992

*Unless otherwise indicated the values are linked to March prices in the respective years

The rows highlighted in grey are the poverty lines linked to the two data points that are analysed in this report. The findings relate to the application of these poverty lines to survey data collected through the Living Conditions Survey (LCS) 2008/09 and LCS 2014/15. For the LCS 2008/09, the survey data and respective poverty lines have been benchmarked to March 2009 prices as this was the mid-point of the survey. However, for the LCS 2014/15 the survey mid-point was April 2015 and therefore the survey data and poverty lines have been benchmarked to that point in time. Relative poverty refers to poverty defined in relation to other elements in the population. This method requires ranking of the elements (households or individuals) according to a welfare indicator, either income or expenditure, from the highest to the lowest. A relative poverty line would therefore be set at say 30% of median household income or expenditure. People or households at the bottom rung, i.e. 30%, will then be referred to as poor. Stats SA has conducted analysis on relative poverty through classifying households into deciles and quintiles and profiling them mostly using money-metric data collected from the IESs and LCSs published in their relevant statistical releases. In 2014, Statistics South Africa published the first South African Multidimensional Poverty Index (SAMPI). The SAMPI was constructed based on the Alkire-Foster method using data from Censuses 2001 and 2011. The

5

Alkire-Foster method focused on four dimensions, namely, health, education, living conditions and economic activity. These deprivation profiles were analysed to identify who is poor and then used to construct a multidimensional poverty index (MPI). The South African Multidimensional Poverty Index (SAMPI) is an adaptation of the global MPI for the South African context. In 2016 an updated SAMPI was released based on the Community Survey (CS) conducted in 2016. Just as the MPI, the SAMPI focused on non-money metric measures of poverty; for instance, lack of education, no access to clean drinkable water or electricity, unemployment etc. Subjective poverty, as mentioned earlier, refers to the perceptions of individuals or households on what constitutes a socially acceptable standard of living in a society from which they make judgements on whether they view themselves as poor or not. In South Africa, most studies on poverty have focused on the objective income measurement, however, similar with elsewhere in the world it has also been recognised that this narrow focus on money-metric measurement of poverty is not adequate and that poverty is influenced by more than just income. Posel and Rogan (2013) argue that the narrow money-metric measurement of poverty in South Africa does not capture the benefits of the social assistance programmes and the pro-poor expenditures that have been advanced by the government since the fall of apartheid on the provision of services such as access to basic services, improved healthcare, education access and housing (i.e. termed ‘social wage’) which may have an influence on how households view themselves subjectively. Unlike publications on the measurement of objective poverty, research on subjective poverty measurement in South Africa is still very recent and the first subjective well-being data in the country was only collected in 1993 where it was presented in the report on the Project for Statistics on Living Standards and Development (PSLD) (Jansen et al, 2013). The first report on subjective poverty in South Africa was released by Stats SA in 2012 and this current report builds on that report. The report Statistics South Africa released in 2012 on subjective poverty was entitled “Subjective Poverty in South Africa”. The report profiled subjective poverty in South Africa using data collected through the Living Conditions Survey (LCS) conducted in 2008/09. The main purpose of this current report is to provide an updated subjective poverty profile of South Africa using the most recent LCS data collected in 2014/15.

6

1.3 Objectives of the report

There are four main objectives of this report. The first is to investigate the level and profile of subjective poverty in South Africa. The second one is to compare subjective poverty levels of 2009 to those of 2015 whilst the third objective is to compare subjective poverty levels to objective poverty levels of the country. The last objective is to analyse the drivers of subjective poverty. The report seeks to answer four main questions, i.e.:

• What are the current levels of subjective poverty in South Africa? • What is the direction of subjective poverty levels in South Africa when comparing 2009 and 2015 data

collection points? • How do subjective poverty levels compare to objective poverty levels? • What drives subjective poverty?

1.4 Methodology and data sources

This section discusses the methodologies used to collect data on the 2008/09 and 2014/15 surveys as they are data sources of this report 1.4.1 Data sources

There are two main data sources that are used in this report, namely, the LCS 2008/09 and LCS 2014/15. LCS was conceptualised in 2007 by Stats SA in order to fill the poverty data gap in the country. This survey was designed to measure poverty in South Africa using various definitions. Both Living Conditions Surveys collected information on income, consumption and spending patterns of households. Similar data collection methods were employed in both surveys which are diary and recall methods. This methodology was first used in the 2005/06 Income and Expenditure Survey (IES) and was later refined in surveys that followed. The improvements included reducing the number of modules in the household questionnaire from 7 modules to 4 modules and also reducing the time for diary keeping from 4 weeks to 2 weeks. This was done mainly to avoid respondents’ fatigue, which may result in respondents pulling out of the survey. LCS 2008/09 data files that were used for producing this report had 25 075 households and 97 486 persons and the LCS 2014/15 data files had 23 080 households and 88 906 persons. SAS statistical software package was used for analysing the data.

7

The data processing of the LCS 2008/09 and LCS 2014/15 involved the processing of completed instruments. These included household questionnaires, diaries and summary questionnaires. Information received from these instruments collected during fieldwork was converted into electronic format represented by numbers or characters. The main method used for this conversion was scanning. To ensure quality electronic data, the data were verified, as well as edited and checked for consistency according to the predetermined editing rules. Decision table technology was used to develop edit rules that were compiled to produce SAS edit programs for the purpose of identifying logical, consistency and out of range errors. In the case of missing key demographic variables (age, sex and population group), hot-deck imputation was used. Hot deck imputation was also used to impute missing money-metric data. 1.4.2 Methods

LCSs of 2008/09 and 2014/15 contained three questions on subjective poverty measures which will be used to estimate the levels of subjective poverty in South Africa (see Table 1.2). How the following subjective poverty measures were constructed was presented in the “Subjective Poverty in South Africa” report in 2012: In the LCS data, the Self-Perceived Wealth Question (SPWQ) asks respondents to identify the category which they consider to best describe the wealth status of their households. Response items are constructed in ordinal scale ranging from ‘very poor’ to ‘wealthy’. Those who responded ‘poor’ or ‘very poor’ are identified as ‘poor’ and all other categories are classified as ‘non-poor’. A subjective poverty line following the Leyden approach, [named for its origin at Leyden University in the Netherlands in the 1970s (Van Praag and Frijters 1999; Ravillion 2012)], is constructed through a minimum income question (MIQ) which asks respondents to select the smallest level of income with which their household could make ends meet. If reported per capita household consumption falls below this minimum income level then the household (and all individuals living in it) are identified as poor. The advantage to this method is that the extent, depth and severity of poverty can be estimated. For example, using the standard Foster-Greer-Thorbecke (FGT) set of poverty measures (Foster et al. 1984), the average distance of each household from its reported minimum income (as a proportion of this level of income) can be estimated (i.e. the depth of poverty). A direct way to use the minimum income question available in the LCS is to ask respondents whether or not their households’ actual level of income is above or below the minimum level reported in the previous question. In this way, respondents evaluate their own perceptions on whether they receive more than their reported minimum level (IEQ). In the LCS, the response items are reported in an ordinal scale ranging from ‘much lower’ to ‘much higher’. Individuals are therefore identified as ‘poor’ if they live in a household in which income is described as ‘lower’ or ‘much lower’ than the

8

minimum required income. All other responses are identified as ‘non-poor’. Logit regressions are used to determine the determinants of subjective poverty for each of the three indicators.

Table 1.2: Subjective poverty indicators available in the LCS 2008/09 and LCS 2014/15

Subjective poverty indicator Response items Poverty cut-off Self-perceived wealth question (SPWQ) “Would you say your household is at present …

1 = Wealth 2 = Very comfortable 3 = Reasonably comfortable 4 =Just getting along 5 = Poor 6 = Very poor

5 = Poor 6 = Very poor

Minimum income question (MIQ) “Which net household income per month in Rand would be the absolute minimum for your household? That is to say, that you would not be able to make ends meet if you earned less.”

Continuous If reported per capita household consumption falls below the perceived minimum income level then the household and its individuals are classified as poor.

Income evaluation question (IEQ) “Is the total monthly income of your household higher, lower or more or less the same as this figure (i.e. the figure given in Q22.9)?”

1 = Much higher 2 = Higher 3 = More or less the same 4 = Lower 5 = Much lower

4 = Lower 5 = Much lower

1.5 Limitations of the report

The subjective poverty questions highlighted in Section 1.4.2 were all asked at household level. The responses to the questions were provided by only one person within the household who was responding on behalf of other household members. This approach assumes that all members of a household will have the same perception about their economic well-being and thus will be classified as per the perception of the person responding on behalf of the household. This assumption may not be true for all instances as perceptions may be influenced by age and sex. It is, however, an advantage that both the LCS 2008/09 and LCS 2014/15 use the same questions and both asked them at household level. For comparability with the 2012 report, analysis in this report is done at individual level, unless otherwise stated.

9

1.6 Layout of the report

This report has seven chapters. The remaining chapters are organised as follows: Chapter 2 provides the key findings of the report. Chapter 3 presents subjective poverty profile based on self-perceived wealth indicator. Chapter 4 presents subjective poverty profile based on minimum income indicator. Chapter 5 presents subjective poverty profile based on income evaluation indicator. Chapter 6 presents findings on living circumstances Chapter 7 provides summary and conclusions

10

Chapter 2Key findings

Chapter 2Key findings

13

This chapter presents the key findings of the report. There are four key findings and they are based on the SPWQ indicator. These four key findings are discussed in sub-sections 2.1, 2.2, 2.3 and 2.4 of this chapter. 2.1 Decline in subjective poverty in South Africa between 2009 and 2015

Table 2.1 compares proportions of subjective poverty in 2009 and in 2015 at both individual and household levels. The results indicate that nationally, the percentage of people who lived in a household perceived as poor has declined from 39,2% in 2009 to 35,0% in 2015. The table also shows that the percentage of households perceived as poor has also declined from 37,6% in 2009 to 33,8% in 2015. This same pattern (decline in subjective poverty) is found among black Africans and coloureds whereas a different pattern is observed among Indians/Asians and white headed households. Table 2.1: Proportion of individuals and households who perceived themselves as poor in 2009 and 2015 Unit of analysis 2009 2015

Individuals 39,2 35,0

Households 37,6 33,8

Figure 2.1 presents the proportion of subjective poverty by population group. The figure shows that the proportion of black Africans and coloureds who see themselves as poor has declined between 2009 and 2015, the proportion of black Africans declined from 46,6% to 40,8% respectively while that of coloureds declined from 19,8% to 18,3% respectively. The proportion of Indian/Asians who see themselves as poor remained the same at 8,6% while the proportion of whites who perceive themselves as poor increased from 3,5% to 4,6% between 2009 and 2015. This proportion of black Africans who see themselves as poor declined by 5,8 percentage points, which is the biggest decline in all the population groups. Coloureds (1,5 percentage point) showed the smallest decrease. Indian/Asians (0,2 percentage point) showed the smallest increase which can be interpreted as negligible, while the proportion of whites who perceived themselves as poor increased by 1,1 percentage point.

14

46,6

19,8

8,4

3,5

39,2 40,8

18,3

8,6

4,6

35,0

0

10

20

30

40

50

Black African Coloured Indian/Asian White South Africa

2009 2015

Figure 2.1: Proportion of subjective poverty by population group; 2009 and 2015

2.2 A comparison of subjective poverty levels and objective income poverty levels in 2009 and 2015

Table 2.2 compares subjective poverty levels using the Self-perceived Wealth Question (SPWQ) indicator and objective income poverty levels using the (Lower Bound Poverty Lines) LBPL at the individual level. The LBPL is the poverty line which is mostly used throughout the report when comparing the subjective and objective poverty levels, it was chosen because it is the line that has been adopted in the National Development Plan by the national government to ensure that objective poverty is eradicated by 2030. Declines in poverty, whether based on subjective poverty or objective income poverty, are observed between 2009 and 2015. Whilst the same pattern is found for both poverty measures, it is also observed that lower proportion of the population perceive themselves as poor compared to proportion who are classified objectively as poor using income measures of poverty.

15

Table 2.2: Subjective and objective poverty headcount rates; 2009 and 2015 Poverty indicator 2009 2015

SPWQ 39,2 35,0

LBPL 47,6 40,0

Table 2.2 indicates that between 2009 and 2015 objective income poverty headcount rates using the lower-bound poverty line declined from 47,6% in 2009 to 40,0% in 2015, similarly to a decline in subjective poverty headcount rates from 39,2% in 2009 to 35,0% in 2015. However, the table also highlights that the subjective poverty headcount rates are lower than those of objective income poverty during the same period, 35,0% and 40,0% respectively in 2015.

2.3 Similarities between subjective poverty and objective poverty in 2009 and 2015

Table 2.3 presents overlaps between subjective poverty and objective income poverty. The results indicate that the majority of the people classified objectively as poor using income poverty measures also perceive themselves as poor. About five out of ten people who were classified objectively as poor also perceived themselves as poor in both 2009 (57,9%) and 2015 (55,3%). Table 2.3: Distribution of individuals by poverty status based on objective and subjective poverty measures; 2009 and 2015

Subjective poverty measure

2009 2015

LBPL LBPL

SPWQ 57,9 55,3

16

2.4 Socially perceived necessities and subjective poverty status

Table 2.4: Proportion of individuals in households with access to socially perceived necessities by subjective poverty status; 2015

Socially perceived necessities RSA Non-poor Poor

1. Meat, fish or vegetarian equivalent every day 62,9 73,6 43,0

2. Special meal on festive events, such as Christmas or equivalent 76,7 84,1 62,9

3. Clothing sufficient to keep your household members warm and dry 79,5 88,5 62,6

4. Some new (not 2nd hand or handed-down) clothes 67,1 75,3 51,8

5. School uniforms for children (if you have children) 69,8 70,4 68,5

6. Pay or contribute to funerals/funeral insurance/ burial society 71,1 74,8 64,3

7. Regular savings for emergencies 32,4 42,5 13,6

8. A house that is strong enough to stand up to the weather e.g. rain, winds, etc. 70,2 80,0 52,0

9. Separate bedrooms for adults and children 67,9 74,5 55,7

10. A bath or shower in the house 41,2 53,5 18,5

11. Burglar bars installed in the house 46,6 57,7 26,0

12. A garden 37,8 41,7 30,7

13. A fence or wall around the property 70,7 77,4 58,1 Table 2.4 shows the proportion of individuals that have the socially perceived necessities by subjective poverty status in 2015. The table indicates that overall a smaller percentage of poor individuals have each of the socially perceived necessities when compared to those that are non-poor. For instance, 62,6% of individuals that are poor have sufficient clothing to keep them warm and dry. However, this percentage is smaller than that of non-poor individuals where 88,5% have sufficient clothing to keep warm and dry. What the table also indicates is that only 13,6% of individuals have regular saving for emergencies whilst 42,5% of non-poor individuals have regular savings for emergencies. Only about 52,0% of poor individuals lived in houses that are strong enough to stand up to the harsh weather conditions compared to 80,0% of non-poor individuals. It is also presented in the table that a significantly smaller percentage of poor individuals live in houses that have a bath or shower where only 18,5% of individuals live in houses with a bath or shower. However, their percentage is considerably smaller than that of individuals that are non-poor where 53,5% of such individuals lived in houses with a bath or shower. Also illustrated from the table is that only 26,0% of poor individuals lived in houses that have burglar bars installed in their houses. Again compared to individuals that are non-poor their percentage is far smaller

17

when compared to non-poor individuals where 57,7% lived in houses that have burglar bars installed. It can also be seen from the table that for a far smaller percentage of poor individuals there is a fence or wall around their property where these accounted for only 58,1% compared to 77,4% of non-poor individuals where there is a fence or wall around their property. Table 2.5: Proportion of individuals in neighbourhood with access to socially perceived necessities by subjective poverty status; 2015

Socially perceived necessities RSA Non-poor Poor

1. Tarred roads close to the house 58,9 67,4 43,1

2. A place of worship (church/mosque/synagogue) in the local area 92,9 94,1 90,7

3. A neighbourhood without rubbish/refuse/garbage in the streets 57,3 60,7 51,0

4. A neighbourhood without smog/smoke in the air 56,8 59,1 52,6

5. Police on the streets in the local area 53,7 59,6 42,9

6. A large supermarket in the local area 52,9 60,1 39,5

7. Somewhere for children to play safely outside the house 38,5 45,0 26,5

8. Street lights 55,5 63,2 41,3

Table 2.5 indicates that poor individuals lived in neighbourhoods that had less access to all the socially perceived necessities compared to those that are non-poor. The table shows that 43,1% of poor individuals lived in neighbourhoods that had access to tarred roads close to their houses compared to 67,4% of the non-poor individuals. Only 42,9% of poor individuals lived in neighbourhoods where police are visible on the streets in their local area, this is considerably lower compared to non-poor individuals where 59,6% lived in neighbourhoods where there is visible police presence on the streets in their local area. Compared to non-poor individuals a smaller percentage of poor individuals lived in neighbourhoods that had access to a large supermarket in their local area, with 39,5% having access compared to 60,1% of non-poor individuals. The table also shows that 26,5% of poor individuals lived in neighbourhoods where children had somewhere to play safely outside the house. This proportion is considerably smaller when compared to non-poor individuals where 45,0% of individuals lived in neighbourhoods where children had somewhere safe to play outside the house. A far lower proportion of poor individuals lived in neighbourhoods that have streets lights where 41,3% lived in neighbourhood that had street lights compared to 63,2% of non-poor individuals.

18

Chapter 3Self-perceived wealth indicator

Chapter 3Self-perceived wealth indicator

21

This chapter profiles subjective poverty in South Africa using the SPWQ indicator. It is sub-divided into three sections. Section 3.1 looks at the demographic characteristics and geographic location of individuals in households that perceive themselves as poor. Section 3.2 compares subjective poverty levels to objective income poverty levels in 2015. Lastly, Section 3.3 explores determinants of subjective poverty. Subjective poverty in this chapter refers to households/individuals who perceive themselves as poor or very poor in the self-perceived wealth question.

3.1 Profiling subjective poverty using the self-perceived wealth question in South Africa; 2009 and 2015

Table 3.1 below shows that in 2015 out of 23 336 households and 88 647 individuals in those households, 35,0% of individuals are from households that perceived themselves as poor. Black Africans had the highest proportion of those who perceived themselves as poor (40,8%) followed by coloureds (18,3%), and then Indians/Asians (8,6%). Whites had the lowest percentage of individuals who are from households that perceived themselves as poor in 2015 at 3,5%. An overall decrease in the percentage of individuals from households who perceive themselves as subjectively poor is observed between 2009 and 2015, 39,2% and 35,0% respectively. The decrease is also observed among black Africans and coloureds. A different pattern is observed among Indians/Asians where their proportion remained the same while for whites an increase in subjective poverty is observed.

22

Table 3.1: Proportion of subjectively poor (SPWQ) by population group; 2009 and 2015 Population group 2009 2015

Black African 46,6 40,8

Coloured 19,8 18,3

Indian/Asian 8,4 8,6

White 3,5 4,6

South Africa 39,2 35,0

The results in Table 3.2 indicate that Eastern Cape, Limpopo and North West provinces had the highest proportion of those that are subjectively poor: with proportions of 52,3%, 45,5% and 41,5% respectively in 2015. On the other hand, Gauteng and Western Cape had the lowest proportions of those that are subjectively poor: 25,7% and 22,7% respectively in 2015. All the provinces, except for Mpumalanga experienced a decrease in the proportion of individuals that viewed themselves as poor or their proportion of individuals that are poor remained the same between 2009 and 2015. The proportion of individuals that viewed themselves as poor from Mpumalanga increased from 34,6% to 36,1% between 2009 and 2015, respectively. Interestingly, in terms of objective poverty Mpumalanga is the only province that experienced a decrease in its poverty head-count between 2011 and 2015. However, of the provinces that experienced a decrease in the proportion of those that viewed themselves as poor KwaZulu-Natal and Free State, experienced the largest percentage point decrease between 2009 and 2015.

23

Table 3.2: Proportion of subjectively poor (SPWQ) by province; 2009 and 2015

Province 2009 2015

Western Cape 22,7 22,7

Eastern Cape 54,3 52,3

Northern Cape 38,7 36,5

Free State 42,1 35,9

KwaZulu-Natal 43,2 33,9

North West 47,3 41,5

Gauteng 28,8 25,7

Mpumalanga 34,6 36,1

Limpopo 50,8 45,5


Table 3.2 also shows that provinces that are predominantly rural registered the largest proportions of those that are poor compared to predominantly urban provinces such as Gauteng and Western Cape. This finding is confirmed in Table 3.3 below where higher proportions of those that are poor are found in traditional areas (48,0%) followed by urban informal areas (47,1%), and then rural formal areas (42,1%) in 2015. The lowest levels of subjective poverty are found in urban formal areas (25,1%) in 2015. The table also indicates that subjective poverty levels declined in all types of settlements between 2009 and 2015. With the exception of the urban formal areas, all the other settlement types, their subjective poverty levels between 2009 and 2015 were above the national averages of 39,2% and 35,0% respectively.

24

Table 3.3: Proportion of subjectively poor (SPWQ) by type of settlement; 2009 and 2015 Type of settlement 2009 2015

Urban formal 27,9 25,1

Urban informal 50,0 47,1

Traditional area 53,9 48,0

Rural formal 48,8 42,1


Table 3.4: Proportion of subjectively poor (SPWQ) by age and sex of head of household; 2015

Age of household head

Sex of household head

Male Female <35 31,4 41,7

35 - 54 26,6 40,5

55 - 64 31,3 45,0

65+ 33,3 47,0


Table 3.4 shows that in 2015 individuals in female-headed households had by far the highest proportion of those that are poor at 43,0% compared with the proportion of individuals in male-headed households which had only 29,2% proportion of poor individuals. This confirms the finding from the broader poverty literature that female headed households are more likely to be poor than male headed households. Both household types reveal that individuals in households headed by a person between the ages of 35 and 54 years old had the lowest proportion of those that are poor at 26,6% for individuals in male-headed households and 40,5% for individuals in female-headed households. The poverty risk between the two household types in this 35 and 54 years old age group is far higher for those in female-headed households with its higher poverty levels compared to those in male-headed households. According to the two household types, individuals in households headed by a person aged 65 years and older have the lower proportion of those that are poor at 33,3% for those in male-headed households compared to those in female-headed households which had 47,0%. Again individuals in female-headed households have by far the largest poverty levels for the 65 years and older age group compared to those in male-headed households.

25

3.2 Comparing subjective poverty levels (SPWQ) and objective income poverty levels; 2015

Section 3.2 compares the poverty headcount of subjective poverty levels and objective income poverty levels in 2015. The comparisons are mainly focusing on the Self-perceived Wealth Question (SPWQ) indicator to the LBPL, however, Figure 3.1 compares subjective poverty levels (SPWQ) to all three income poverty lines, namely, FPL, LBPL and UBPL. Figure 3.1: Poverty headcount by poverty line used; 2015

Figure 3.1 shows the poverty headcount for three objective poverty lines and subjective poverty indicator SPWQ for 2015. The figure indicates that the objective poverty headcounts when compared to the poverty headcount of the SPWQ indicator are generally higher except for the FPL (25,2%) which is considerably lower than the SPWQ indicator (35,0%) in 2015. The UBPL (55,5%) which shows the highest poverty headcount of all the objective poverty measures is considerably higher than the poverty headcount of SPWQ indicator. The LBPL at 40,0% is the closest of the objective poverty measures to the poverty headcount of the SPWQ indicator although it’s still a bit higher than that of the subjective measure.

25,2

40,0

55,5

SPWQ 35,0

0

10

20

30

40

50

60

FPL LBPL UBPL

26

Table 3.5: Percentage distribution of households by poverty status based on the LBPL line and the subjective poverty measure; 2015

LBPL

SPWQ

Poor Non-poor

Non-poor 57,7 83,7

Poor 42,3 16,3

Total 100,0 100,0

Table 3.5 above compares the poverty status of households between the subjective poverty measure (SPWQ) and the LBPL in 2015. The table shows that there is considerable overlap between the subjective poverty status and the objective poverty status where it reveals that 42,3% of households that viewed themselves as subjectively poor were also classified as poor according to the LBPL. The table also reveals that the SPWQ indicator when compared to the objective measure identifies different households as being poor. Table 3.6: Percentage share of poverty by province and poverty indicator; 2015 Province LBPL SPWQ RSA

KwaZulu-Natal 26,0 19,2 19,9

Eastern Cape 18,6 18,8 12,6

Limpopo 14,9 13,5 10,4

Gauteng 11,4 17,6 24,0

North West 7,9 8,0 6,7

Mpumalanga 8,3 8,1 7,8

Western Cape 6,0 7,3 11,3

Free State 4,7 5,3 5,1

Northern Cape 2,2 2,2 2,2

South Africa 100,0 100,0 100,0

Table 3.6 shows that in 2015 KwaZulu-Natal province had the largest share of individuals living below the LBPL (26,0%) and they were also identified as having the largest share of those that are subjectively poor (19,2%), however, its share of those that are poor is lower than that of the objective measure. Eastern Cape was

27

measured as having the second largest share of individuals living below the LBPL at 18,6%. The Eastern Cape is also identified as the province with the second largest share of those that are subjectively poor at 18,8%. Limpopo was measured as the province with the third largest share of individuals living below the LBPL and those that are subjectively poor at 14,9% and 13,5% respectively. For KwaZulu-Natal, Eastern Cape and Limpopo provinces, their share of individuals living below the LBPL and those that are subjectively poor were over-represented because they were higher than the relative overall population share in their respective provinces. For example KwaZulu-Natal, according to the objective poverty measure; 26,0% of its population is living below the LBPL which is more than its share of the total population of 19,9%. Northern Cape had the lowest share of those living below the LBPL and those who are subjectively poor where both measures had the same share of 2,2%. Gauteng and Western Cape provinces were under-represented in terms of their share of individuals living below the LBPL and subjectively poor relative to their share of the total population of South Africa. Gauteng had a total population share of 24,0% yet its share of the individuals living below the LBPL and subjectively poor were 11,4% and 17,6% respectively, and these are smaller than the province’s population share. However, the reason why Gauteng has the largest share of poor individuals is mainly as a result of the large population size of the province. Table 3.7: Percentage share of poverty by settlement type and poverty indicator; 2015 Settlement type LBPL SPWQ RSA

Urban formal 29,6 39,7 55,3

Urban informal 10,7 11,1 8,2

Traditional area 54,7 44,5 32,6

Rural formal 5,0 4,7 3,9

South Africa 100,0 100,0 100,0

28

Table 3.7 indicates that according to the LBPL and SPWQ indicator, traditional areas had the largest share of individuals living below these respective poverty lines in 2015 where their shares were 54,7% and 44,5% respectively. However, the shares of those that are poor according to the SPWQ indicator is considerably lower than that of the LBPL. Both poverty measures identify individuals residing in urban formal areas as having the second largest shares of those that are poor at 29,6% according to the LBPL and 39,7% according to the SPWQ indicator, where the share of the SPWQ indicator is considerably higher than that of the LBPL. Individuals residing in traditional areas were over-represented in terms of their share of the poor where their share of individuals living below the LBPL and are subjectively poor was higher than their share of the total population of 32,6%. Both poverty measures identified individuals in rural formal areas as having the lowest share of individuals living below LBPL and SPWQ indicator at 5,0% and 4,7% respectively. However, individuals residing in rural areas were over-represented in terms of their share of the poor relative to their total population share where they had a total population share of 3,9%, although their share of individuals living below the LBPL and SPWQ indicator were higher than their share of the total population. Table 3.8: Percentage share of poverty by population group and poverty indicator; 2015

Population group LBPL SPWQ RSA

Black African 94,7 93,7 80,4

Coloured 5,1 4,6 8,8

Indian/Asian 0,1 0,6 2,5

White 0,1 1,1 8,3

South Africa 100,0 100,0 100,0

Table 3.8 indicates that in 2015 black African population group had the largest share of individuals that were poor according LBPL and SPWQ indicator and that their share of the poor were over-represented relative to their total population. The shares of the black African population that are poor according to the LBPL and SPWQ indicator were 94,7% and 93,7% respectively. The coloured population group had the second largest share of those that are poor according to the LBPL and SPWQ indicator at 5,1% and 4,6% respectively and they were underrepresented in terms of their share of the poor relative to their total population. Whites and Indians had the lowest shares of those that are poor according to both poverty measures where they were also underrepresented in terms of their share of the poor relative to their total population.

29

3.3 Subjective poverty profile using Socially Perceived Wealth Indicator (SPWQ)

Poverty estimates by the highest education level of the household head are illustrated in Table 3.9. The table shows that individuals in households headed by a person with no schooling had the highest poverty levels at 58,2% in 2009 and 54,5% in 2015 respectively, which is way above the national averages of 39,2% in 2009 and 35,0% in 2015. Table 3.9: Poverty headcount rates by highest education level of household head; 2009 and 2015 Education level 2009 2015

No schooling 58,2 54,5

Incomplete primary 54,5 50,7

Completed primary 48,0 42,9

Incomplete secondary 37,7 36,2

Matric 18,5 18,6

Tertiary 5,4 4,9


Individuals in households headed by a person with incomplete primary education as their highest level of education had the second highest level of poverty where in 2009, 54,5% viewed themselves as poor and in 2015 it decreased to 50,7%. The table also indicates that the higher the educational attainment of the household head the less likely for that household to view itself as poor. Individuals in households headed by a person with a tertiary qualification as their highest educational level had the lowest proportion of those that viewed themselves as poor between 2009 (5,4%) and 2015 (4,9%) respectively. 3.3.1 Individuals identified as subjectively poor but measured as objectively non-poor.

Table 3.10 illustrates the percentage distribution of individuals and households that viewed themselves as poor but at the same time were measured as objectively non-poor using the lower-bound poverty line.

30

Table 3.10: Percentage distribution of individuals and households classified as subjectively poor but measured as objectively non-poor; 2009 and 2015 2009 2015



It is indicated from the table that in 2009 at the individual level 29,8% of individuals lived in households that self-reported as subjectively poor while they were measured as objectively non-poor, but this figure increased to 37,0% in 2015. At the household level there was an increase in the percentage of households that viewed themselves as poor but were also measured as objectively non-poor between 2009 and 2015 where these increased from 48,7% to 57,7%, respectively. Table 3.11: Percentage share of individuals classified as subjectively poor but measured as objectively non-poor, by socially perceived necessity of regular saving for emergencies; 2009 and 2015

Regular savings for emergencies 2009 2015

Have 17,8 21,0

Don't have cannot afford 74,6 75,8

Don't have and don't want 4,4 2,3

Other 3,2 0,9

Total 100,0 100,0 Table 3.11 presents the share of individuals who viewed themselves as poor and were also measured as objectively non-poor disaggregated by the socially perceived necessity of regular savings for emergencies. The table shows an interesting pattern between 2009 and 2015 which might help to explain why some households might have viewed themselves as poor yet have been measured as objectively non-poor. It is shown in the table that between 2009 and 2015 the largest share of individuals resided in households which reported that according to this socially perceived necessity they could not afford to regularly save for emergencies, in 2009 this share was 74,6% and in 2015 it increased to 75,8%. Individuals that reported that they are able to regularly save for emergencies were 17,8% in 2009 and they rose to 21,0% in 2015. Individuals

31

that reported that they intentionally did not want to regularly save for emergencies had a share of 4,4% in 2009 and this decreased to 2,3% in 2015. In 2009 and 2015 the other category represents those individuals that resided in households that reported that they either did not know or whose answer was that the asked question was not applicable to them, or who simply did not respond to the question during data collection, the share of these individuals were 3,2% and 0,9% in 2009 and 2015 respectively. Thus for individuals who resided in households that responded that they could not afford to save could be due to various factors such as that the households might be living in debt, they might be supporting extended family members living elsewhere, etc. but are able to make enough ends to meet to live above the LBPL. Thus due to their inability to save for emergencies they might view themselves as feeling poor because they are financially vulnerable because if something happens to the household that requires immediate spending of money, they might not be able to pay for such an event. Table 3.12: Percentage share of individuals classified as subjectively poor but measured as objectively non-poor, by socially perceived necessity of having a bath or shower in the house; 2009 and 2015

Bath or shower in the house 2009 2015

Have 28,5 33,8


Don't have and don't want 4,8 -

Other 4,4 3,7

Total 100,0 100,0

The share of individuals residing in households that are subjectively poor and were also measured as objectively non-poor, disaggregated by the socially perceived necessity of having a bath or shower in the house is reported in Table 3.12. The table provides further evidence of why individuals who are residing in households that were subjectively poor yet were measured as objectively non-poor using the LBPL might have viewed themselves as such. It is shown by the table that in both 2009 and 2015 the majority of individuals resided in households that reported that according to the socially perceived necessity of having a bath or shower in their houses they did not have and also could not afford to have either. They had a share of 62,4% in 2009 and this remained the same at 62,5% in 2015. Therefore by not having a bath or shower in their houses because they could not afford one, such households might regard their housing standards as not being adequate which might have contributed in viewing themselves as poor because they are perhaps using other rooms in their houses that

32

were not designed as bathrooms or shower rooms and when say perhaps they want to wash themselves, they might not have any privacy. Individuals that reported that they resided in households that have a bath or shower were 28,5% in 2009 and they increased to 33,8% in 2015. Those that reported that they did not have and also did not want a bath or shower were 4,8% in 2009 and in 2015 no one responded affirmatively to this category. The other category was 4,4% in 2009 and it decreased to 3,7% in 2015. Table 3.13: Percentage share of individuals classified as subjectively poor but measured as objectively non-poor, by socially perceived necessity of having somewhere for children to play safely outside the house; 2009 and 2015

Somewhere for children to play safely outside the house 2009 2015

Have or have access 28,6 34,9

Don't have 58,2 63,4

Other 13,3 1,7

Total 100,0 100,0 Table 3.13 indicates the share of individuals who are subjectively poor and were also measured as objectively non-poor, disaggregated by the socially perceived necessity of having somewhere for children to play safely outside the house. Another indication of why individuals residing in households that are subjectively poor yet were measured as objectively non-poor could have viewed themselves as such is illustrated by the table. From the table it is seen that in 2009, 58,2% of individuals resided in households which did not have this socially perceived necessity of having somewhere safely for children to play outside the house, however, in 2015 an even higher share of individuals reported that they did not have this socially perceived necessity and it increased to 63,4%. Not having a safe place for children outside the house to play might have influenced households’ perceptions on how they viewed themselves subjectively where they might have felt that they are poor subjectively because their children cannot play safely outside the house yet they were measured to live above the LBPL. The share of individuals that indicated they have or had access outside for their children to play safely was 28,6% in 2009 and this share saw an increase in 2015 to 34,9%. The other response category includes those individuals which resided in households that reported that they did not know how to respond to the asked question, those which felt that this question was not applicable to them and also those that did not respond to the question during data collection, and the share of this category was 13,3% in 2009 and in 2015 it decreased to 1,7%.

33

3.3.2 Individuals classified as subjectively non-poor but measured as objectively poor

Table 3.14: Percentage distribution of individuals and households classified as subjectively non-poor but measured as objectively poor; 2009 and 2015 2009 2015



The percentage distribution of individuals and households classified as subjectively non-poor but measured as objectively poor is shown in Table 3.14. The table shows that in 2009, 32,9% of individuals stayed in households which were subjectively non-poor but were measured as objectively poor, using the lower-bound poverty line but in 2015 witnessed a decline of such households to 27,4%. The share of households which are subjectively poor but were measured as objectively non-poor was 20,6% in 2009 and this declined to 16,3% in 2015. Table 3.15: Percentage share of individuals classified as subjectively non-poor but measured as objectively poor, by socially perceived necessity of regular saving for emergencies; 2009 and 2015 Regular savings for emergencies 2009 2015

Have 16,6 23,0


Don't have and don't want 4,7 1,4

Other 5,0 0,6

Total 100,0 100,0

Table 3.15 shows that individuals that resided in households that were subjectively non-poor and also measured as objectively poor using LBPL disaggregated by the socially perceived necessity of regular savings for emergencies. The table shows that in both 2009 and 2015 the share of those individuals who could not afford to make regular savings for emergencies were the highest. The individuals who could not afford to make regular savings for emergencies were 73,7% in 2009 and this share rose to 74,9% in 2015. Individuals that indicated that they were making regular savings for emergencies had a share of 16,6% in 2009 and this increased to 23,0% in 2015. Individuals who indicated that they did not make regular savings for emergencies and they did not want to, had a share of 4,7% in 2009 and in 2015 their share declined to 1,4%. Individuals who responded to

34

the other category are those that indicated that they either did not know the answer or the asked question was not applicable to them or that the question was not responded to during data collection, had a share of 5,0% in 2009 and this share declined to 0,6% in 2015.

Table 3.16: Percentage share of individuals classified as subjectively non-poor but measured as objectively poor, by socially perceived necessity of having a bath or shower in the house; 2009 and 2015 Bath or shower in the house 2009 2015

Have 14,0 19,7


Don't have don't want 6,0 2,7

Other 7,1 0,5

Total 100,0 100,0

Table 3.16 shows that individuals that resided in households that were subjectively poor and also were measured as objectively non-poor using the LBPL disaggregated by the socially perceived necessity of having a bath or shower in the house. Those with the largest share of 72,9% in 2009 could not afford to have this socially perceived necessity and their share increased to 77,1% in 2015. For individuals that reported that they have a bath or shower in their house, their share increased from 14,0% in 2009 to 19,7% in 2015. In 2009 a share of 6,0% individuals indicated that they did not have and did not want this socially perceived necessity in their house, their share dropped to 2,7% in 2015. Those that reported other for this socially perceived necessity had a share of 7,1% in 2009, however, their share declined to 0,5% in 2015. Table 3.17: Percentage share of individuals classified as subjectively non-poor but measured as objectively poor, by socially perceived necessity of having somewhere for children to play safely outside the house; 2009 and 2015 Somewhere for children to play safely outside the house 2009 2015

Have or have access 25,4 26,6

Don't have 66,6 72,9

Other 8,1 0,5

Total 100,0 100,0

35

The percentage share of individuals that were subjectively non-poor and were also measured as objectively poor using the LBPL according to the socially perceived necessity of having somewhere for children to play safely outside the house are shown in Table 3.17. The table shows that in both 2009 and 2015, individuals that resided in households that indicated that they did not have this socially perceived necessity had the highest share between 2009 and 2015. In 2009 their share was 66,4% and 2015 saw an increase of their share to 72,9%. The individuals that reported that they had access to this socially perceived necessity also experienced an increase in their share between 2009 and 2015 of 25,4% and 26,6% respectively. The other category was reported by those individuals who resided in households that reported that they did not know the response to the asked question, those which felt that this question was not applicable to them and as well as those that did not respond to the question, where in 2009 this share was 8,1% and in 2015 this share declined to 0,5%.

3.4 Determinants of self-perceived wealth indicator

Table A.1 in Annexure A presents the logit regression coefficients which estimate the likelihood that an individual is living in a household that perceives itself to be poor (according to the SPWQ indicator). The coefficients presented in these regressions denote the natural logarithm of the odds of being poor. The regression yield data for 88 647 individuals after observations with missing values for key variables are dropped. Compared with the other population groups, black Africans are more likely to be poor. Individuals in urban informal areas are more likely to be poor compared to those in urban formal areas. Residing in urban formal and rural formal areas are associated with the risk of poverty compared to residing in urban formal area. However, residing in traditional area offers a protective risk against poverty compared to residing in an urban formal area. Individuals in Eastern Cape, Northern Cape, Free State, North West, and Limpopo are more likely to be poor compared to those in the Western Cape. However, individuals in KwaZulu-Natal, Gauteng and Mpumalanga are less likely to be poor compared to those in Western Cape. Larger households have a small likelihood of being poor. Having a higher number of children is associated with less risk of being poor. Having a greater number of employed persons in a household offers a protective effect against poverty. Having access to flush toilet, piped water and electricity and having a household refuse removed by a local municipality are associated with a lower risk of poverty. Living in a household where the household head is married or is cohabiting offers a strong significant protective effect against poverty. Living in a dwelling where there is a wall or fence around the property is associated with lower risk of poverty. Having a garden offers protective effective against poverty. Being injured in the past month and having needed medical treatment is associated with a higher risk of poverty. Being a victim of crime is associated with the risk of

36

poverty. A higher proportion of male household members is associated with a higher risk of poverty. Living in a formal dwelling offers a strong protective effect against poverty. A dwelling unit that is owned by a household member offers protective effect against poverty for those living in it.

Chapter 4Minimum income indicator

Chapter 4Minimum income indicator

39

This chapter outlines subjective poverty in South Africa based on the Minimum Income Question (MIQ) measure. The MIQ measure is constructed by asking respondents to indicate the household’s minimum level of income without which the household could not survive and compares this minimum income threshold with the total per capita household consumption that they report. In the event that their per capita household consumption is less than the minimum income threshold, the household and everyone who is part of it is considered to be poor. The MIQ approach allows the possibility to estimate the poverty gap (i.e. the average distance poor households are from the poverty line) and severity of poverty (i.e. takes into account not only how far households are from the poverty line but also the inequality among the poor) using a subjective method because it is the only approach that uses both a poverty threshold (minimum income) and the reported per capita household consumption (total household expenditure). This chapter consists of three sections where section 4.1 explores the demographic characteristics and geographic location of individuals in households who are considered to be poor based on the MIQ measure. Section 4.2 compares subjective poverty (MIQ) levels to objective income levels for 2015 as well as explore the depth and severity of poverty from a subjective (MIQ) perspective. Lastly, in section 4.3 we explore the factors that are correlated with subjective poverty.

4.1 Profile of subjective poverty (MIQ) in South Africa; 2009 and 2015

Table 4.1 indicates that 59,3% of the population was considered to be poor in 2009 using the MIQ subjective poverty measure and that percentage decreased to 50,7% in 2015. The decrease is observed across all the different population groups with the greatest drop being noted for Indians/Asians which fell from 55,0% in 2009 to 28,3% in 2015. Even though there was an overall decrease, the percentage of black Africans that are subjectively poor (52,6%) is still higher than the national estimate which is 50,7%. Based on this measure, in 2009 black Africans had the highest incidence of subjective poverty (61,8%), followed by Indians/Asians (55,0%) and coloureds coming in at 52,6%. The white population group had the smallest prevalence at 45,0%. The percentage of individuals who are poor declined from 59,3% in 2009 to 50,7% in 2015.

40

Table 4.1: Proportion of subjectively poor (MIQ) by population group; 2009 and 2015 Population group 2009 2015


Coloured 52,6 48,9


White 45,0 41,0


Table 4.2 indicates that in 2015 the Eastern Cape and Northern Cape had the highest percentage of individuals that were identified as poor at 70,2% followed by the Free State at 67,8%. During the same period, KwaZulu-Natal and Gauteng had the lowest percentages of those identified as poor with 27,1% and 46,1% respectively. Between 2009 and 2015, there has been a general decrease in the percentage of poor individuals except in the Western Cape, Northern Cape and Limpopo. Among all those provinces that showed a reduction in poverty, the biggest decrease was observed in KwaZulu-Natal which dropped by nearly 50% from 60,4% in 2009 to 27,1% in 2015. In the Western Cape, Northern Cape and Limpopo, there was an increase in the percentage of poor persons. These are the only provinces where subjective poverty was observed to have increased. Table 4.2: Proportion of subjectively poor (MIQ) by province; 2009 and 2015 Province 2009 2015







Gauteng 56,7 46,1


Limpopo 49,5 63,3


41

Table 4.3 demonstrates that subjective poverty levels based on the MIQ indicator have declined for all the different settlement types between 2009 and 2015, with the biggest decrease was observed in urban formal areas from 56,5% to 47,7%. Even though there has been an overall reduction across all settlement types during the period under review, the percentage of poor individuals have remained above the national average in both years except for those residing in urban formal areas. Table 4.3: Proportion of subjectively poor (MIQ) by type of settlement; 2009 and 2015 Type of settlement 2009 2015




Traditional 61,1 54,6


Table 4.3 indicates that in 2015 the highest prevalence of subjective poverty was among households in urban informal settlements (55,2%); followed by traditional areas (54,6%) and rural formal areas (51,0%). The lowest prevalence of subjective poverty was found in the urban formal areas (47,7%), where most of the economic opportunities exist. Table 4.4 shows that overall, individuals in female-headed households who are classified as poor compared to their male counterparts had the lowest proportion. Individuals in female-headed households where the household head’s age is less than 35 years showed a smaller proportion (almost half of those in male-headed households) of individuals who are classified as poor (19,3%). This is in contrast to their male counterparts who show a higher percentage of those considered to be poor at 36,1%, which is almost twice that of female headed households.

42

Table 4.4: Proportion of subjectively poor (MIQ) by age and sex of head of household; 2015



Male Female

<35 36,1 19,3

35 – 54 31,8 19,9

55 – 64 25,7 21,8

65+ 20,7 26,7


This pattern continues for female headed households where the age of the household head is between the ages of 35 and 54, resulting in 19,9% of them being regarded as poor compared to their male counterparts at 31,8%. Where the household head’s age is between 55 and 64 years, a similar story can be told. However, when the household head is over the age of 65 and male, the proportion considered to be poor is 20,7% compared to their female counterparts at 26,7%. An interesting finding illustrated by Table 4.4 is that as the male-headed households get older the percentage of individuals identified as poor reduces while at the same time, as the female-headed households get older the proportion of individuals experiencing poverty increases.

4.2 Comparing subjective poverty levels (MIQ) and objective income poverty levels; 2015

Section 4.2 examines subjective poverty levels based on the MIQ measure and objective income poverty levels in 2015 with the emphasis on the MIQ estimate and the lower-bound poverty line (LBPL), however, Figure 4.1 compares subjective poverty levels (MIQ) to all three income poverty lines, namely, FPL, LBPL and UBPL.

43

Figure 4.1: Poverty headcount by poverty line used; 2015

Figure 4.1 shows the poverty headcount for the subjective poverty indicator (MIQ) and the three objective poverty lines for 2015. When comparing the objective and subjective headcount rates, the MIQ indicator (50,7%) is higher except for the UBPL which is 55,5%, which also has the highest poverty headcount for all the objective poverty measures. When comparing the LBPL to the subjective indicator, the figure shows that subjective poverty using the MIQ indicator is higher at 50,7%; compared to the objective poverty measure that stands at 40,0%.

25,2

40,0

55,5

MIQ 50,7

0

10

20

30

40

50

60

FPL LBPL UBPL

44

Table 4.5: Poverty indicators by province; 2015

Province

LBPL Subjective poverty (MIQ)

Poverty gap (%) Severity of poverty Poverty gap (%) Severity of poverty

Western Cape 6,8 3,0 23,1 13,2

Eastern Cape 27,5 15,8 34,4 21,2

Northern Cape 15,9 8,2 34,7 21,2

Free State 13,5 6,7 32,4 19,4

KwaZulu-Natal 22,5 12,6 10,5 5,3

North West 19,5 10,6 25,3 15,0

Gauteng 6,2 2,9 20,3 11,8

Mpumalanga 16,9 8,8 26,2 15,9

Limpopo 26,6 15,6 30,6 18,5

South Africa 16,6 9,1 23,2 13,6

Table 4.5 shows that the poverty gap and severity of poverty tend to be higher for provinces who have a higher poverty headcount. Stated differently, the average distance of poor individuals from their respective poverty lines is greater for those provinces in which a larger percentage of individuals are below the poverty line. The fact that individuals are poor based on the objective poverty line does not mean that they are poor based on the subjective MIQ measure. For example, the Northern Cape was the fourth highest province based on the poverty gap from the LBPL (15,9%) while Gauteng (6,2%) and the Western Cape (6,8%) had the smallest poverty gaps of all the provinces. On the other hand, the Northern Cape (34,7%) is the poorest province based on the subjective poverty estimates resulting from the MIQ method with the Eastern Cape (34,4%) following a close second. The poor residents of KwaZulu-Natal reported a shortfall of 10,5% from their minimum income levels which it is also the lowest poverty gap based on the subjective measure of all the provinces and about half of the national average (23,2%) while at the same time being the third poorest in terms of the LBPL (22,5%).

45

Table 4.6: Poverty estimates by settlement type; 2015

Settlement type



Urban formal 7,4 3,6 20,9 11,9

Urban informal 19,3 9,8 28,2 17,8

Traditional areas 31,2 18,1 25,9 15,4

Rural formal 20,4 10,8 24,5 14,4

South Africa 16,6 9,1 23,2 13,6

Table 4.6 presents the estimates of the poverty gap and severity of poverty based on the subjective poverty line represented by MIQ indicator as well as the objective LBPL. Based on the LBPL, the table suggests that poor individuals in traditional areas have the biggest gap (31,2%) from the objective poverty line compared to the other settlement types. On the other hand, the subjective poverty gap for poor people in urban informal areas (28,2%) is greatest, followed by traditional areas (25,9%) and rural formal areas at 24,5%. The depth of poverty for poor individuals living in urban formal areas is smallest for both the objective (7,4%) and the subjective (20,9%) measure, which is also lower than the national average at 16,6% and 23,2% respectively. The table suggests that traditional area dwellers are living the furthest below the objective poverty line while the MIQ measure suggests that urban informal dwellers are the ones that are furthest away. The severity of poverty estimates has a positive relationship with the poverty gap estimates since the severity of poverty increases as the poverty gap increases.

46

Table 4.7: Poverty estimates by sex of the household head; 2015 LBPL Subjective poverty (MIQ)

Sex of the household head Poverty gap (%) Severity of poverty Poverty gap (%) Severity of poverty

Male 12,4 6,6 23,1 13,6

Female 22,5 12,5 23,4 13,6

South Africa 16,6 9,1 23,2 13,6

The disparities in terms of the severity and depth of poverty between males and females are presented by Table 4.7, using both the objective as well as the subjective measures. Based on the LBPL, the average distance from the poverty line (22,5%) is bigger for individuals living in female-headed households relative to those with a male head (12,4%). These estimates are higher and lower to the national average of 16,6%. According to Table 4.7, the poverty gap estimates using the MIQ measure suggest that those individuals living in a household with a female head (23,4%) are further below their minimum income poverty lines relative to individuals living in male-headed households (23,1%). The table shows that a smaller fraction of individuals in male-headed households are living below their minimum incomes with the severity of poverty indicator identical for both sexes as well as the country as a whole at 13,6%. Comparing the objective and subjective measure, it is interesting to note that the proportion of poor individuals in female-headed households (22,5%) is almost twice that of poor individuals in male-headed households (12,4%), based on the LBPL. In the case of the subjective measure, the estimates for male-headed and female-headed households are nearly identical at 23,1% and 23,4% respectively. This is an indication that the nature of poverty among males and females in South Africa is changing, based on the MIQ measure.

47

Table 4.8: Poverty estimates by population group; 2015

Population group



Black African 19,8 10,9 24,9 14,8

Coloured 8,3 3,9 20,2 11,1

Indian/Asian 0,3 0,1 9,4 4,3

White 0,1 0,1 14,4 7,4

South Africa 16,6 9,1 23,2 13,6

As shown in Table 4.8, the subjective poverty gap for poor black Africans (24,9%) is the largest, meaning they are furthest away from their minimum incomes compared to the other population groups. Poor people that are part of the coloured population report a shortfall of 20,2% from their minimum income line; based on the MIQ measure but 8,3% on average below the lower-bound poverty line. The shortfall for whites from their minimum income level is estimated at 14,4% using the MIQ indicator while they are only 0,1% below the objective lower-bound poverty line. Poor Indian/Asians are the closest to their minimum income level (9,4%) accompanied by the lowest severity of poverty (4,3%) which is also lower than the national average while at the same time being only 0,3% below the objective poverty line. Based on the table, it is clear that huge disparities exist between the objective and subjective measure of poverty, especially in the case of the Indian/Asian and white population groups. Table 4.9: Distribution of households by poverty status based on the lower-bound poverty line and subjective poverty measure; 2015

LBPL

MIQ

Poor Non-poor

Non-poor 67,8 82,8

Poor 32,2 17,2

Total 100,0 100,0

Table 4.9 above examines the poverty status of households between the subjective poverty estimate (MIQ) and the LBPL in 2015. The table illustrates that the intersection between the subjectively derived poverty status

48

(MIQ) and the objectively derived poverty status shows that 32,2% of households identified as poor based on the MIQ subjective poverty indicator were also measured as poor using the LBPL. The table also shows that these two different poverty measurements characterise different households as being poor. Table 4.10: Percentage share of poverty by province and MIQ poverty indicator; 2015 Province LBPL MIQ RSA







Gauteng 11,4 21,9 24,0


Limpopo 14,9 13,0 10,4

South Africa 100,0 100,0 100,0

Table 4.10 shows that in 2015, KwaZulu-Natal had the largest share of individuals living below the LBPL at 26,0%, followed by the Eastern Cape and Limpopo at 18,6% and 14,9% respectively. Gauteng had the largest share of individuals identified as subjectively poor in 2015 at 21,9%. Eastern Cape province had a similar picture for both those measured as objectively poor as well as subjectively poor at 18,6% and 17,4% respectively, which makes the province the second largest with a share of poor individuals irrespective of poverty measurement used. The movements are similar for Limpopo which makes it the province with the third largest share regardless of the poverty indicator used. The share of individuals living below the LBPL were understated in the Western Cape and Free State because their share was less than the overall population share for those provinces. For instance, the Western Cape’s share of the total population is 11,3% while the share of individuals living below the LBPL is 6,0%. Similarly, the share of the poor in Gauteng province were understated in terms of both the objective poverty measure and the subjective poverty measure compared to the total share of the Gauteng population. The table illustrates that the total share of the Gauteng province is 24,0% whereas its share of individuals living below the LBPL and the MIQ measure is 11,4% and 21,9% respectively. On the other hand, in the Eastern Cape, North West, Mpumalanga and Limpopo the share of individuals classified as poor

49

using the objective measure as well as those using the subjective measure were overrepresented in relation to the total share of the population in their respective provinces. For example, Limpopo’s total share is 10,4% whereas its shares of individuals living below the LBPL and the MIQ indicator are 14,9% and 13,0% respectively. There were no over- or under-representations in the Northern Cape, except for the MIQ measure that showed overrepresentation. Table 4.11: Percentage share of poverty by settlement type and MIQ poverty indicator; 2015 Settlement type LBPL MIQ RSA



Traditional areas 54,7 35,1 32,6


South Africa 100,0 100,0 100,0

Table 4.11 shows that based on the objective poverty line (LBPL) the greatest share of the poor were living in traditional areas (54,7%) while individuals in urban formal areas had the second largest share at 29,6%. On the other hand, if we use the subjective poverty approach, the second biggest share of the poor live in traditional areas (35,1%) with urban formal dwellers having the largest share at 52,0%. In traditional areas the share of the subjectively poor (35,1%) is about one and a half times less than those classified as objectively poor (54,7%), while the subjective measure (52,0%) in urban formal areas are almost twice the size of the objective measure (29,6%). Traditional areas were overrepresented in terms of their shares of the poor based on both the LBPL (54,7%) and the MIQ indicator (35,1%) in that it was higher than the share of the total population (32,6%) living in that particular settlement type. According to the table, both poverty measures show that the lowest share of individuals classified as objectively poor (5,0%) or subjectively poor (3,9%) occurred in rural formal areas. Contrary to traditional areas, the residents of urban formal areas were underrepresented based on both the objective poverty measure (29,6%) and the subjective poverty measure (52,0%) which was lower than the share of the total population (55,3%) residing in urban formal areas.

50

Table 4.12: Percentage share of poverty by population group and poverty indicator; 2015 Population group LBPL MIQ RSA




White 0,1 6,7 8,3

South Africa 100,0 100,0 100,0

Table 4.12 shows that the black African population group in 2015 represented 94,7% of individuals who were classified as objectively poor based on the LBPL and 83,4% of individuals who were identified as subjectively poor using the MIQ indicator. The table also indicates that the members of this particular population group were overrepresented in terms of their share of the poor, based on the LBPL (94,7%) and the MIQ (83,4%) measures because they were higher than the total share of the population (80,4%) that belongs to this population group. Coloured people had the second largest share of those that were classified as both objectively (5,1%) and subjectively (8,5%) poor, and contrary to black Africans, they were underrepresented in terms of their share of the poor compared to the total coloured population (8,8%). Indians/Asians and whites had the lowest share of individuals who are both objectively and subjectively poor. These two population groups were underrepresented in terms of their share of the poor relative to the total of their respective population group. Interestingly among the white population, there are 67 times more individuals who are classified as subjectively poor (6,7%) compared to those who are considered to be objectively poor (0,1%).

51

4.3 Determinants of minimum income indicator

Table A.1 in Annexure A presents the logit regression coefficients which estimate the likelihood that an individual is living in a household identified as poor (according to the MIQ indicator). The coefficients presented in these regressions denote the natural logarithm of the odds of being poor. The regression yield data for 88 906 individuals. Compared with the other population groups, black Africans are more likely to be poor. Residents of Eastern Cape, Northern Cape, Free State and Limpopo are more likely to be poor compared to Western Cape residents. On the other hand, those individuals from the KwaZulu-Natal, North West, Gauteng and Mpumalanga provinces are coupled with a lower risk of being poor relative to those in the Western Cape. Living in rural formal areas provides a protective effect against poverty compared to those living in urban formal areas. Living in urban informal areas and traditional areas compared to those living in urban formal areas is associated with the risk of poverty The table shows that the bigger the household an individual is part of, the lower the likelihood that the individual will experience poverty while the number of children in the household has a positive association with living in a household that is poor. The higher number of employed household members offers a protective effect against poverty. Living in a dwelling unit that has a flush toilet and access to electricity is associated with lower risk of poverty, however, those in a dwelling unit with piped water and whose refuse is removed by a local municipality are associated with a higher risk of being poor. Living in a household where the household head is married or is cohabiting is associated with a higher risk of poverty. Living in a dwelling where there is a wall or fence around the property and there is also a garden is associated with lower risk of poverty. Being injured in the past month and having needed medical treatment is associated with the risk of poverty, however, being a victim of crime is associated with the risk of poverty. Having a higher proportion of household members who are male is significantly associated with the higher risk of poverty. Living in a formal dwelling unit which is owned by a household member offers a significant protective effect against poverty.

52

Chapter 5Income evaluation indicator

Chapter 5Income evaluation indicator

55

This chapter profiles subjective poverty in South Africa using the IEQ indicator. It is sub-divided into three sections. section 5.1 looks at the demographic characteristics and geographic location of individuals in households that were identified as subjectively poor. Section 5.2 compares subjective poverty levels to objective income poverty levels of 2015. Lastly, section 5.3 explores determinants of subjective poverty using the IEQ indicator. For this chapter subjective poverty is determined at the household level where the income evaluation question was asked to one person within the household who then evaluated their own perceptions and responded on behalf of the household that the household’s actual income was ‘lower’ or ‘much lower’ than the reported minimum income level that the household cannot live without. However, the results unless otherwise stated are reported at the individual level where if a household identified itself as poor every member within that household will have the same poverty status. Therefore, in this chapter, subjective poverty refers to households whose total income is less than the reported income perceived to be the absolute minimum by households.

5.1 Profile of subjective poverty (IEQ) in South Africa; 2009 and 2015

Table 5.1 presents that in 2015 out of 22 698 households and 86 598 individuals in those households, 50,4% of individuals are from households that are poor. In 2015, black Africans had the highest proportion of those who are poor (54,1%) followed by coloureds (43,9%), and Indians/Asians who had the third highest proportion at 28,9%. Whites had the lowest proportion at 26,2% of individuals that are poor. Table 5.1: Proportion of subjectively poor (IEQ) by population group; 2009 and 2015 Population group 2009 2015


Coloured 43,8 43,9


White 28,0 26,2


Table 5.1 also highlights that between 2009 and 2015 there was an overall decrease in the proportion of individuals that are poor from 58,0% to 50,4%. Except for the coloured population group whose proportion of those that are poor remained the same, all the other population groups experienced a decrease in their

56

proportions of those that are poor between 2009 and 2015. However, the Indian/Asian population group had the highest percentage point decrease of the proportion of those that are poor followed by the black African population group. Between 2009 and 2015 black Africans were the only population group whose proportion of those that are poor was above the national averages. Table 5.2 shows that the in 2015, the Eastern Cape (72,8%) had the highest proportion of those who identified themselves as poor compared to Northern Cape (66,1%) which had the second highest proportion of the poor, followed by Limpopo (64,1%). KwaZulu-Natal and Gauteng had the lowest proportion of those that identified themselves as poor at 33,2% and 39,1% respectively, in 2015. Western Cape, Northern Cape, Mpumalanga and Limpopo are the only provinces that experienced an increase in the proportions of those that are poor between 2009 and 2015. However, of the provinces that experienced an increase in the proportion of those that are poor Limpopo had the highest percentage point increase followed by the Western Cape. Eastern Cape, Free State, KwaZulu-Natal, Northern Cape and Gauteng are the provinces that experienced a decrease in the proportion of those that are poor between 2009 and 2015. However, KwaZulu-Natal, which was one of the provinces that experienced a decline in their proportion of those that are poor, had the highest percentage point decrease. Western Cape and Gauteng provinces are the only provinces in 2009 and 2015 where their proportion of those that are poor was always below the national averages of 58,0% and 50,4% respectively. Table 5.2: Proportions of subjectively poor (IEQ) by province; 2009 and 2015 Province 2009 2015







Gauteng 50,8 39,1


Limpopo 44,1 64,8


57

Table 5.3 confirms the findings presented in Table 5.2 which show that provinces which are predominantly rural had the highest proportions of those that are poor. Individuals in traditional (61,1%) and rural formal (49,4%) areas in 2015 had the highest proportion of those that are subjectively poor. Table 5.3: Proportion of subjectively poor (IEQ) by type of settlement; 2009 and 2015 Type of settlement 2009 2015



Traditional area 66,7 61,1



Individuals in urban informal areas (46,7%) had the third highest proportion of those that are poor in 2015, and whilst individuals in urban formal areas (44,5%) had the lowest proportion. The table also shows that between 2009 and 2015 the proportion of those that are poor declined in all settlement types. However, individuals in urban informal areas experienced the highest percentage point decrease. The decline in poverty levels was minimal (single digit decline) in rural formal and traditional areas compared to urban formal and urban informal areas. In 2015, it is only individuals in traditional areas where the proportion of those that are poor were higher than the national average (50,4%). Between 2009 and 2015 the only settlement type where the proportion of those that are poor were below the national averages of 58,0% and 50,4% respectively were the urban formal and rural formal areas. Table 5.4 shows that in 2015 more than half (54,7%) of individuals in female headed-households were poor, whereas less than half (47,3%) of individuals living in male-headed households were poor.

58

Table 5.4: Proportion of subjectively poor (IEQ) by age and sex of head of household; 2015



Male Female

<35 48,3 55,3

35 - 54 45,5 54,8

55 - 64 49,5 54,0

65+ 49,4 54,5

Total 47,3 54,7

Generally individuals in female-headed households had the highest prevalence of subjective poverty for all age groups considered compared to their male counterparts. Households where the female head is aged less than 35 years old had the highest proportion of those that are poor (55,3%). This was considerably higher than their male counterparts at 48,3%. Individuals from households headed by males aged between 35 and 54 years old had a lower proportion of those who are poor (45,5%) followed by those that are aged less than 35 years old (48,3%). Individuals from households headed by males aged between 55 and 64 years old and those aged 65 years and above had the highest proportion of those that are poor at 49,5%. In female-headed households individuals with the highest proportion of those that are poor are from households headed by persons aged less than 35 years old with a proportion of 55,3%. This is different from individuals in male-headed households where those with the highest proportion are from the age group 35 and 54 years old. Overall the lowest proportion for poor female headed households was for those aged 55 to 64 years (54,0%) albeit it was still quite high when compared to the male headed households in the same age group. However, in the male-headed households this group has the highest proportion of individuals that are poor (49,5%).

5.2 Comparing subjective poverty levels (IEQ) and objective income poverty levels; 2015

Section 5.2 compares subjective poverty levels to objective income poverty levels in 2015. The comparisons are mainly focusing on the IEQ indicator to the LBPL, however, Figure 5.1 compares subjective poverty levels (IEQ) to all three income poverty lines, namely, FPL, LBPL and UBPL.

59

Figure 5.1: Poverty headcount rates by poverty line used; 2015

Figure 5.1 presents the poverty headcount rates based on the three objective national poverty lines and the subjective measure. The figure indicates that the poverty headcount rates of the IEQ indicator (50,4%) are higher than two of the three objective poverty lines FPL (25,2%) and LBPL (40,0%) respectively but lower than the UBPL (55,5%). The IEQ is much closer to the UBPL, which might indicate that the households’ perceptions of poverty may be broader (when one considers food and non-food items) than just considering the inadequacy of food. Table 5.5: Distribution of households by poverty status based on the lower-bound poverty line and the subjective poverty measure; 2015

LBPL

IEQ

Poor Non-poor

Non-poor 68,9 79,8

Poor 31,1 20,2

Total 100,0 100,0

25,2

40,0

55,5

IEQ 50,4

0

10

20

30

40

50

60

FPL LBPL UBPL

60

Tables 5.5 above compares the poverty status of households between the subjective poverty measure (IEQ) and the LBPL in 2015. It is indicated from the table that there are similarities between the subjective and objective poverty measures where 31,1% households that have been classified as subjectively poor were also measured as objectively poor using LBPL. The table also reveals that apart from identifying more households as poor compared to the LBPL the subjective indicator (IEQ) also identifies different households as being poor.

Table 5.6: Percentage share of poverty by province and poverty indicator; 2015 Province LBPL IEQ RSA



Limpopo 14,9 13,6 10,4

Gauteng 11,4 18,3 24,0






South Africa 100,0 100,0 100,0

Table 5.6 shows that in 2015 according to the LBPL, KwaZulu-Natal was measured as the province with the largest share of individuals living below LBPL at 26,0%, Eastern Cape was measured as the province with the second largest share of individuals below the LBPL at 18,6% and Limpopo was measured as the province with the third largest share of individuals living below the LBPL at 14,9%. Northern Cape was measured as the province with the smallest share of individuals living below the LBPL at 2,2%; followed by Free State at 4,7%. KwaZulu-Natal, Eastern Cape, Limpopo, North West and Mpumalanga province were overrepresented in terms of their share of the poor relative to their total population share because their share of the poor is more than the total population of their respective provinces. For example, KwaZulu-Natal has 19,9% of the total population but has 26,0% share of all poor South Africans. Western Cape, Gauteng and Free State provinces were underrepresented in terms of their shares of the poor because these provinces’ respective total populations are less than their respective share of the poor. For example, Western Cape has 11,3% of the overall population but only has 6,0% of the poor.

61

According to the IEQ indicator it identified Gauteng and Eastern Cape as the provinces with the joint largest share of individuals which are poor at 18,3%, respectively and this differs with the objective poverty measure which identified KwaZulu-Natal as the province with the largest share of the poor. Limpopo was identified as the province with the second largest share of individuals which are subjectively poor at 13,6%; followed by KwaZulu-Natal with a share of 13,1%. Similarly with the objective measure, Northern Cape was identified as the province with the smallest share of individuals that are subjectively poor at 2,9%. Eastern Cape, Limpopo, North West, Mpumalanga, Free State and Northern Cape were overrepresented in terms of their shares of the poor relative to their respective proportions of their total populations. Gauteng was identified as the province with the joint largest share of the poor and this was largely as a result of its large population of 24,0% of the country’s population. Even though Gauteng had the largest population share, it was still underrepresented in terms of its share of the poor. Other provinces that were underrepresented in terms of their share of the poor relative to their total population are KwaZulu-Natal and Western Cape. Table 5.7: Percentage share of poverty by settlement type and poverty indicator; 2015 Settlement type LBPL IEQ RSA



Traditional areas 54,7 40,2 32,6


South Africa 100,0 100,0 100,0

Table 5.7 shows that according to the LBPL, individuals in traditional areas (54,7%) had the largest share of those living below the poverty line. Individuals in urban formal areas (29,6%) had the second largest share of those living below the LBPL. Individuals in rural formal areas (5,0%) had the smallest share of those living below the LBPL. Individuals residing in urban informal, traditional and rural formal areas were overrepresented in terms of their shares of the poor relative to their overall population. For example the proportion of the total population of individuals residing in traditional areas is 32,6% yet their share of the poor of (54,7%) is higher than their overall population. Individuals residing in urban formal areas were underrepresented in terms of their share of the poor relative to their total population (i.e. the settlement type contained 55,3% of the overall population but only 29,6% of the poor).

62

The IEQ indicator differs with the objective poverty measure where it identifies individuals residing in urban formal areas as having the largest share of those that are poor at 48,2%. It also differs with the objective poverty measure where it identifies individuals in traditional areas (40,2%) as having the second largest share of those that are poor. Similarly, with the LBPL it identifies individuals in rural formal areas as having the smallest share of those that are poor at 3,9%. Individuals in traditional areas are also overrepresented in terms of their share of the poor relative to their overall population according to the IEQ indicator as it was with the LBPL. However, the IEQ indicator differs with the LBPL where it identifies individuals in urban informal areas as underrepresented in terms of their share of the poor whereas the LBPL identified them as overrepresented. Table 5.8: Percentage share of poverty by population group and poverty indicator; 2015 Population group LBPL IEQ RSA




White 0,1 4,1 8,3

South Africa 100,0 100,0 100,0

Table 5.8 indicates that according to the LBPL and IEQ indicator the black African population group had the largest share of those that are poor at 94,7% and 86,8% respectively where the share of the poor when using the LBPL is more than that of the IEQ indicator. The black African population group were overrepresented in terms of their shares of the poor according to both poverty measures where their respective shares of the poor at 94,7% for LBPL and 86,8% for IEQ indicator which are more than the share of their total population (80,4%). Coloured people were similarly identified by both poverty measures as having the second largest shares of those that are poor where the share of the poor according to the IEQ indicator is higher at 7,7% compared to that of LBPL at 5,1%. Both poverty measures were underrepresented in terms of their shares of the poor relative to the proportions of their total population. Indian/Asian and white population groups were identified as having the smallest shares of the poor and they were underrepresented in terms of their share of the poor relative to their overall populations.

63

5.3 Determinants of income evaluation indicator

Table A.1 in Annexure A presents the logit regression coefficients which estimate the likelihood that an individual is living in a household that identified itself to be poor (according to the IEQ indicator). The coefficients presented in these regressions denote the natural logarithm of the odds of being poor. The regression yield data for 86 598 individuals after observations with missing values for key variables are dropped. Compared with the other population groups, black Africans are significantly more likely to be poor. Residing in Eastern Cape, Northern Cape, Free State, Mpumalanga and Limpopo is associated with higher risk of poverty compared to those living in Western Cape. Whereas residing in KwaZulu-Natal, North West and Gauteng offers a protective effect against poverty. Age, household size and number of children in the household are associated with the risk of poverty. A greater number of employed persons in a household offers a protective effect against poverty. Living in a dwelling which uses flush toilet, piped water and in which the refuse is removed by the local municipality is associated with a higher risk of poverty. However, living in a dwelling which has access to electricity is associated with lower risk of poverty. Living in urban informal areas offers a protective effect against poverty compared to living in urban formal areas. However, residing in traditional and rural formal areas is associated with higher risk of poverty. Living in a household where the household head is married or cohabiting offers a protective effect against poverty. Living in a dwelling which has a fence or wall around the property is associated with a smaller likelihood of being poor. Residing in a dwelling that has a garden offers protective effect against poverty. Being injured in the past month and having required medical treatment is associated with the risk of poverty. Living with a victim of crime is associated with a lower risk of poverty. A higher proportion of males in a household is associated with higher risk of poverty. Residing in a formal dwelling offers a significantly protective effect against poverty whereas living in a dwelling that is owned by a household member is associated with the risk of poverty.

64

Chapter 6Living circumstances of households

Chapter 6Living circumstances of households

67

This chapter presents a comparison of the living circumstances of households that have been measured as poor using the LBPL and the three subjective poverty indicators. The purpose of the chapter is to determine if there are any differences in living standards between the households that have been measured as poor by LBPL and those that have been identified as poor subjectively since subjective poverty is likely to be driven by more than just a lack of income. It is likely that given the massive pro-poor government expenditure after apartheid that the government has undertaken components of the social wage such as access to basic services might have an influence on subjective poverty assessments by households. Section 6.1 presents the comparisons of average annual household expenditure according to both objective and subjective classifications of the poor. Section 6.2 compares the living standards of households that were classified as poor objectively against those that have been classified as subjectively poor according to the three subjective poverty measures.

6.1 Household expenditure

Table 6.1 provides a comparison of the average annual household consumption expenditure which is based on the households which have been measured as poor objectively and those that have been identified as subjectively poor in 2015, using the three subjective poverty indicators. According to the table, average household consumption expenditure for households that are poor based on the LBPL is significantly less than the expenditure for poor households based on subjective measures. Table 6.1 indicates that objectively poor households’ annual household consumption expenditure is R26 839 per annum while at the same time pointing out that the expenditure for poor households based on SPWQ indicator is considerably higher with an average expenditure of R40 155 per annum. What is interesting is that these households consider themselves to be poor, but their expenditure is almost double that of objectively measured poor households. The average expenditure of households that are subjectively poor according to the MIQ indicator is R74 242 per annum which is almost three times more than the average expenditure indicated by households below the LBPL (R26 839 per annum). By the same token, the levels of expenditure for households below the LBPL is about a third of the expenditure levels of poor households, based on the IEQ measure which is R72 268 per annum. Overall, the lowest average annual household consumption expenditure occurred in Limpopo (R61 011); followed by North West (R69 192) and the Eastern Cape (R72 390). The highest levels of average household consumption expenditure was experienced in the Western Cape (R163 220 per annum) followed by Gauteng at R140 676 per annum.

68

When average annual household expenditure is disaggregated by sex, an interesting development occurs. Expenditure based on households below the LBPL are almost identical for both male-headed and female-headed households at R27 279 per annum and R26 481 per annum respectively. From the table it is quite clear that male-headed households in terms of expenditure are far better off than those headed by their female counterparts across all the subjective poverty indicators. For example, the expenditure of male-headed households that are poor according to the MIQ indicator is R85 905 compared to R57 139 of their female counterparts. It appears that there are variances between poverty levels based on subjective poverty indicators and poverty levels based on objective poverty lines from a population group point of view. The average annual household expenditure of whites (R21 871) below the LBPL was about 20% lower than their black African counterparts at R26 651. Conversely, the expenditure levels based on subjective indicators for whites are far higher than the expenditure levels of all the other population groups.

69

Table 6.1: Average annual household expenditure among the poor by poverty indicator (in Rand); 2015

LBPL SPWQ MIQ IEQ Total

RSA 26 839 40 155 74 242 72 268 103 293

Province

Western Cape 29 350 48 029 107 877 94 767 163 220

Eastern Cape 24 401 32 791 54 504 47 954 72 390

Northern Cape 27 632 38 521 71 073 62 952 81 258

Free State 26 385 38 790 68 571 68 685 85 298

KwaZulu-Natal 28 278 32 574 49 508 67 809 73 503

North West 25 918 35 239 52 541 60 579 69 192

Gauteng 27 729 52 303 100 956 100 968 140 676

Mpumalanga 29 341 40 367 62 246 71 001 83 517

Limpopo 24 463 34 927 48 131 47 557 61 011

Sex of the household head

Male-headed 27 279 42 000 85 905 82 199 121 363

Female-headed 26 481 38 018 57 139 59 697 77 671

Population group

Black African 26 651 38 249 51 833 58 106 67 828

Coloured 30 751 51 013 88 584 85 629 124 445

Indian/Asian 19 915 94 819 167 470 150 095 195 336

White 21 871 105 722 304 976 278 244 350 937

70

6.2 Living circumstances of households

Figure 6.1: Proportion of households that received a government housing subsidy or RDP house by province and poverty indicator; 2015

Figure 6.1 denotes that 21,2% of all South African households that had consumptions below the LBPL have benefitted from a government housing subsidy or RDP house. The households that have been identified as poor, based on all the three subjective poverty indicators were less likely to have benefited from this government programme that is meant to assist poor households with their housing needs. Based on the SPWQ, MIQ and IEQ indicators, which are respectively: 18,5%, 15,3% and 16,1% of the subjectively poor identified households received some sort of housing support from government.

WesternCape

EasternCape

NorthernCape Free State KwaZulu-

NatalNorthWest Gauteng Mpumalan

ga Limpopo Total

LBPL 17,0 9,9 27,5 36,8 21,5 22,6 28,7 20,8 18,9 21,2SPWQ 15,5 10,0 25,2 31,9 18,1 19,0 19,7 19,2 20,8 18,5MIQ 12,6 10,1 20,9 26,8 12,2 18,6 15,7 13,0 16,5 15,3IEQ 14,9 8,9 24,7 28,5 14,2 17,4 18,4 15,7 13,7 16,1

0

5

10

15

20

25

30

35

71

Figure 6.2: Proportion of households who have access to piped water inside their dwellings or on-site by province and poverty indicator; 2015

Figure 6.2 indicates that households that are subjectively poor are more likely to have access to drinking water inside their dwellings or on-site than those households classified to be below the LBPL. Across the entire country, about 52,7% of households who are objectively poor have access to water inside the dwelling or on-site. The households who are subjectively poor have a greater chance to have access to water than their objectively poor counterparts and this is due to the fact that households report that being subjectively poor is more than about not having money or access to piped water. Households in the Northern Cape that are objectively poor are more likely to have access to piped drinking water than households who consider themselves poor using the SPWQ measure.

WesternCape

EasternCape



ga Limpopo Total

LBPL 78,7 25,3 65,0 84,7 47,4 52,8 82,2 53,1 33,5 52,7SPWQ 75,3 32,2 63,1 86,9 55,3 56,2 88,6 61,1 39,8 61,9MIQ 86,2 40,7 75,2 90,5 66,6 67,2 92,0 69,4 50,8 72,7IEQ 86,1 35,1 72,5 92,7 67,8 65,0 93,1 68,1 47,6 69,7

0

25

50

75

100

72

Figure 6.3: Proportion of households that had access to flush toilets inside their dwellings or on-site by province and poverty indicator; 2015

All subjectively poor identified households based on all the three subjective poverty indicators have a greater chance to have access to flush toilets than those households measured to be poor based on the LBPL. Households identified as subjectively poor according to the MIQ indicator have the highest proportion of households with access to flush toilets. The access to flush toilets for households that are objectively poor is restricted to about 31,6% while access to flush toilets of households based on SPWQ, MIQ and IEQ measures are 46,9%, 60,3% and 56,9% respectively.

WesternCape

EasternCape



ga Limpopo Total

LBPL 91,0 16,9 44,0 60,1 20,5 31,8 61,2 14,8 5,5 31,6SPWQ 87,0 25,4 47,2 63,1 35,5 37,4 77,3 28,2 14,4 46,9MIQ 92,9 33,8 63,9 74,6 48,9 53,9 81,6 40,4 25,5 60,3IEQ 90,9 27,1 61,2 77,3 48,1 53,9 84,9 39,4 23,2 56,9

0

25

50

75

100

73

Figure 6.4: Proportion of households living within a 2km radius of selected facilities by type of facility and poverty indicator; 2015

Based on Figure 6.4, objectively poor households below the LBPL are less likely to reside within two kilometres of a secondary school, primary health care, hospital, post office, welfare office or MPCC. According to the figure, there is not much difference between the subjective and objective indicators on the closeness to public transport, food market and the different types of schools from the household. For example, 81,2% of poor households based on the MIQ measure and 80,9% of poor households based on the IEQ indicator reside within a two-kilometre radius of a food market or shop.

Foodmarket

Publictransport

Pre-primary/pre-school

Primaryschool

Secondary school

Clinic/communitycentre

Hospital

Post officeor postofficeagent

Welfareoffice MPCC

LBPL 79,9 92,0 85,2 81,7 65,6 47,3 15,4 30,6 23,6 33,0SPWQ 79,9 91,9 84,6 80,8 66,7 50,9 18,0 35,6 27,7 38,5MIQ 81,2 91,0 84,5 80,5 67,8 54,4 23,6 43,8 32,9 42,2IEQ 80,9 91,3 84,9 81,0 67,3 53,5 21,7 41,5 32,1 41,9

0

25

50

75

100

74

Figure 6.5: Percentage of households in which a member of the household ‘never goes hungry’ by population group and poverty indicator; 2015

Figure 6.5 suggests that there might be a possible relationship between food security and subjective poverty based on the descriptive statistics mentioned in the figure. Households whose poverty status are determined by the minimum income question are more than likely to indicate that household members never go hungry compared to households’ poverty status that is determined by the LBPL. This is confirmed by 70,8% of objectively poor households stating that household members never go hungry compared to 80,1% and 77,6% of poor households classified using the MIQ and IEQ indicators, respectively.

African Coloured Indian White TotalLBPL 70,7 72,7 79,7 30,9 70,8SPWQ 67,6 70,5 57,9 83,0 67,9MIQ 78,1 85,2 91,1 96,7 80,1IEQ 76,0 85,3 83,3 94,0 77,6

0

25

50

75

100

Chapter 7Summary and conclusion

Chapter 7Summary and conclusion

77

Subjective poverty levels in South Africa have decreased between 2009 and 2015 when they are measured by the three subjective poverty measures namely Self-Perceived Wealth Indicator (SPWQ), Minimum Income Question (MIQ) and Income Evaluation Question (IEQ). According to the SPWQ indicator subjective poverty levels declined from 39,2% in 2009 to 35,0% in 2015. Using the MIQ indicator subjective poverty levels declined from 59,3% in 2009 to 50,7% in 2015 whereas in terms of IEQ indicator subjective poverty levels declined from 58,0% to 50,4% respectively. Subjective poverty levels calculated by population group between 2009 and 2015 display different patterns according to the three subjective poverty measures. However, two consistent patterns across the three subjective poverty measures are that the poverty levels for black African individuals have declined significantly between 2009 and 2015 but they were still the highest of all the population groups across the three subjective indicators. The only thing that is different between the three subjective indicators regarding poverty levels of black Africans is the extent of their decline across them. Firstly, using the SPWQ indicator poverty levels for black Africans declined from 46,6% in 2009 to 40,8% in 2015 whereas according to the MIQ indicator their poverty levels declined from 61,8% in 2009 to 52,6% in 2015, and using the IEQ indicator they declined from 63,6% in 2009 to 54,1% in 2015. Therefore, between 2009 and 2015 the IEQ indicator had the largest percentage point decline in the poverty levels of black Africans. Between 2009 and 2015 the IEQ indicator produced the highest proportion of black African individuals that are subjectively poor compared to the other subjective poverty indicators. The subjective poverty levels for the coloured population group between 2009 and 2015 declined only when identified according to the SPWQ and MIQ indicators but remained the same when identified according to the IEQ indicator at 43,9%. Subjective poverty levels for the Indian/Asian population group display an interesting pattern where, according to the SPWQ indicator, it remained virtually the same at 8,6% between 2009 and 2015. However, according to the MIQ and IEQ indicators their poverty levels declined significantly from 55,0% in 2009 to 28,3% in 2015 when using the MIQ indicator whilst in terms of the IEQ indicator they declined from 45,0% in 2009 to 28,9% in 2015 – with the MIQ indicator showing the largest percentage point decline. With regards to the poverty levels of the white population between 2009 and 2015 the MIQ indicator displays a varying pattern between the three subjective poverty indicators. According to the SPWQ indicator the subjective poverty levels for whites increased between 2009 and 2015 from 3,5% to 4,6%, respectively. When applying the MIQ poverty line they declined from 45,0% in 2009 to 41,0% in 2015 and according to the IEQ indicator their subjective poverty levels declined from 28,0% in 2009 to 26,2% in 2015. When the poverty measures of the three subjective poverty indicators are compared with the objective measure using the LBPL, the subjective poverty measures nationally mainly produce higher levels of poverty. Between 2009 and 2015 of all the three subjective poverty measures, only the SPWQ indicator produced lower

78

poverty levels compared to the LBPL poverty levels of 47,6% in 2009 and 40,0% in 2015. In terms of the poverty levels by population group the poverty measures of the LBPL between 2009 and 2015 when compared with the subjective poverty indicators show that for the black African population their poverty levels were only higher compared to the SPWQ indicator but lower than the MIQ and IEQ indicators. With regards to the coloured, the LBPL poverty levels showed the same pattern as that of the black Africans where they were only higher compared to the SPWQ indicator but lower compared to the other subjective indicators’ poverty levels. Subjective poverty levels by settlement type between 2009 and 2015 declined in all settlements. However, with regards to which settlement type had the largest proportion of those that are subjectively poor, they mostly differ between the three subjective poverty measures. In urban formal areas the subjective poverty indicator that produced the highest poverty levels is the MIQ indicator at 56,5% in 2009 and 47,7% in 2015. In urban formal areas the subjective poverty indicator which produced the highest poverty levels differ between 2009 and 2015 where in 2009 the IEQ indicator produced the highest poverty levels at 68,0% and in 2015 it was the MIQ indicator that produced the highest poverty levels at 55,2%. In traditional areas the highest poverty levels were produced by the IEQ indicator between 2009 and 2015 at 66,7% and 61,1%, respectively. Highest poverty levels in formal areas between 2009 and 2015 were produced by the MIQ indicator where it declined from 66,1% to 51,0% respectively. When all the settlement types poverty levels that are produced by the three subjective poverty indicators are compared with those of the objective poverty measures using LBPL between 2009 and 2015 they show differing comparisons. Between 2009 and 2015 the LBPL poverty levels at 26,7% and 21,4% respectively for urban formal areas were lower than all the three subjective poverty indicators’ poverty levels. However, for the formal areas in 2009 at 63,3% they were only higher than poverty levels of the SPWQ indicator whereas in 2015 at 52,2% they were only higher than the SPWQ and IEQ indicators. For the traditional areas between 2009 and 2015 the poverty levels of the LBPL at 76,7% and 67,1%, respectively had the highest poverty levels when compared to all the subjective poverty indicators. In formal areas between 2009 and 2015 the comparison of the poverty levels of LBPL and the three subjective indicators show that in 2009 the LBPL at 61,1% was only higher than the SPWQ indicator, whereas in 2015 the LBPL produced the highest poverty levels at 51,3%. Between 2009 and 2015 the poverty levels of the Western Cape remained the same at 22,7% throughout the period when measured by SPWQ indicator, however, with regards to the MIQ indicator the poverty levels of the province increased from 47,2% to 52,9%, respectively, for both years. The poverty levels of Western Cape also increased according to the IEQ indicator from 31,4% to 45,7%. The poverty levels of the Eastern Cape when using the SPWQ indicator decreased between 2009 and 2015 from 54,3% to 52,3%, respectively. They also

79

decreased according to the MIQ and IEQ indicators from 78,4% to 70,2% respectively in terms of the MIQ indicator and from 80,3% to 72,8%, respectively when using the IEQ indicator. According to the SPWQ indicator in the Northern Cape between 2009 and 2015 the poverty levels decreased from 38,7% to 40,7%, respectively, however, when using the MIQ indicator the poverty levels increased from 67,5% to 70,2% respectively between 2009 and 2015, similarly with the IEQ indicator the poverty levels for Northern Cape increased from 65,8% to 66,1% between 2009 and 2015 respectively. The SPWQ indicator for Free State decreased from 42,1% in 2009 to 35,9% in 2015. Using both the MIQ and IEQ indicators the poverty levels for Free State between 2009 and 2015 decreased. The SPWQ results showed that poverty levels in KwaZulu-Natal decreased from 43,2% in 2009 to 33,9% in 2015. Using the MIQ indicator, KwaZulu-Natal experienced a massive decline in its poverty levels from a high of 60,4% to a low of 27,1%, and the province also experienced a massive decrease in its poverty levels when using the IEQ indicator, declining from 67,9% to 33,2% between 2009 and 2015 respectively. In terms of the North West province its poverty levels declined from 47,3% to 41,5% between 2009 and 2015 according to the SPWQ indicator whilst according to the MIQ indicator it declined from 61,4% to 53,4% between 2009 and 2015 whereas the IEQ indicator declined from 66,5% to 57,3% between 2009 and 2015. Gauteng province’s poverty levels declined when using the SPWQ indicator from 28,8% in 2009 to 25,7% in 2015. The MIQ indicator's poverty levels for Gauteng declined from 56,7% in 2009 to 46,1% in 2015, however, when using the IEQ indicator the poverty levels decreased from 50,8% to 39,1% between 2009 and 2015. Mpumalanga experienced an increase of its poverty levels when the SPWQ indicator is used from 34,6% in 2009 to 36,1% in 2015. However, the province only experienced a decrease in its poverty levels when the MIQ indicator is applied which decreased from 58,4% to 54,2% between 2009 and 2015. Another increase in Mpumalanga provinces’ poverty levels was experienced when the IEQ indicator is used when it increased from 57,3% to 61,3% between 2009 and 2015. Limpopo’s poverty levels decreased from 50,8% to 45,5% when using the SPWQ indicator between 2009 and 2015. Poverty levels according to MIQ indicator increased massively between 2009 and 2015; increasing from 49,5% to 63,3% between 2009 and 2015 whereas for the IEQ indicator they experienced a significant decrease from 44,1% to 64,8% between 2009 and 2015. Therefore between 2009 and 2015 the MIQ indicator produced the highest poverty levels in Western Cape, Northern Cape, Free State and Gauteng. Whilst in 2009 it only produced the highest poverty levels in Mpumalanga. While the SPWQ indicator produced the highest poverty levels in Limpopo in 2009 only and KwaZulu-Natal only in 2015. The IEQ indicator produced the highest poverty levels in Eastern Cape and North West between 2009 and 2015 whereas they only produced the highest poverty levels in KwaZulu-Natal in 2009 only and also produced the highest poverty levels in Limpopo and Mpumalanga in 2015 only. When comparing

80

the subjective poverty indicators with the LBPL, the LBPL only produced the highest poverty levels against all the subjective poverty indicators in Limpopo in 2009 as a joint highest poverty level ranking with MIQ indicator of 58,4%. It also has the highest poverty levels over all the subjective poverty measures in 2009 in Limpopo with a poverty level of 71,5%. In 2015 the poverty levels for individuals in female-headed households were significantly higher than those of individuals in male-headed households according to the SPWQ and IEQ poverty indicators. However, in terms of the MIQ indicator the subjective poverty levels for individuals in both male and female-headed households were similar at 50,7% respectively. Poverty levels for individuals in both male and female-headed households also showed a differing pattern with regard to the poverty indicator that produced the lowest and highest poverty levels. The SPWQ indicator produced the lowest poverty levels for individuals in male-headed households at 29,2%; followed by IEQ indicator at 47,3% whilst the MIQ indicator produced the highest poverty levels for individuals in male-headed households at 50,7%. For individuals in female-headed households the highest poverty levels were produced by the IEQ indicator at 54,7% the MIQ indicator produced the second highest poverty levels for individuals in female-headed households at 50,7%. The SPWQ indicator produced the lowest poverty levels for individuals in female-headed households at 43,0%. The advantage of the MIQ indicator is that it is a money-metric question which has a poverty threshold which can be compared to the reported household per capita consumption to produce a poverty gap and severity of poverty which can then be compared with those of the objective poverty measure. The MIQ indicator when comparing its depth and severity of poverty which declined between 2009 and 2015 with those of the objective poverty measure using the FPL, LBPL and UBPL, as a result of its higher poverty headcount compared with FPL and LBPL but lower against UBPL between 2009 and 2015, its depth of 29,3% and 23,2% between 2009 and 2015 and its severity of 17,9% and 13,6% between 2009 and 2015 are higher than the depth and severity of poverty of the FPL and LBPL. However, they are lower than the depth of 33,5% and 21,3% and severity of poverty of 23,2% and 13,6% of the UBPL between 2009 and 2015. When the poverty status of the three subjective poverty indicators are compared with that of the LBPL they show that there is a considerable overlap between households that have been measured as poor objectively and those that have identified themselves as subjectively poor. In 2009 the percentage of households that were measured as living below the LBPL and those that viewed themselves as subjectively poor according to the self-perceived wealth question (SPWQ) were 51,4% and the percentage of those households that are subjectively poor according to the MIQ and IEQ indicators and also living below the LBPL were 37,3% and 39,1% respectively. In 2015 the percentage of households that were classified as living below the LBPL and also viewed

81

themselves as poor according to the SPWQ indicator were 42,3% which are lower when compared to those households in 2009. Using the MIQ and IEQ indicators in 2015 the percentages of households identified as subjectively poor and also measured as poor when using the LBPL were 32,2% and 31,1% respectively, where they declined between 2009 and 2015. Therefore one can conclude that objective poverty is a best predictor of poverty status based on self-perceived wealth than poverty rates based on minimum income question. It can also be concluded that the subjective poverty measures identify different households as being poor compared to those identified as poor by the objective measure. The logit regressions for the three subjective poverty indicators revealed differing drivers of subjective poverty for each indicator. In terms of all the subjective poverty indicators, individuals that are black African are more likely to be poor compared to all the other population groups. Residing in urban informal areas only offers a protective effect against being poor compared to those in urban formal areas in terms of the IEQ. With regards to SPWQ and MIQ, residing in urban informal areas is associated with the higher risk of being poor. Living in traditional areas only offers a protective effect against poverty compared to those in urban formal according to the SPWQ indicator, according to the MIQ and IEQ living in traditional areas is associated with the risk of poverty. Residing in rural formal areas only offers a protective effect against poverty according to the MIQ indicator, while according to the SPWQ and IEQ indicator residing in rural formal areas is associated with the risk of poverty. In terms of provinces living in Eastern Cape, Northern Cape, Free State and Limpopo you are more likely to be poor according to the three subjective poverty indicators when compared with those of Western Cape. However, residing in KwaZulu-Natal and Gauteng offers a protective effect against poverty according to all the three subjective poverty indicators. Living in Mpumalanga you are more likely to be poor only according to the IEQ indicator, whereas according to the SPWQ and MIQ indicators you are less likely to be poor. Living in North West you are more likely to be poor according to the SPWQ and IEQ indicators only, whereas according to the MIQ indicator you are less likely to be poor. Living in a larger household only offers a protective effect against poverty according to the MIQ indicator, according to the SPWQ and IEQ indicators living in a larger household is associated with the risk of poverty. Having a greater number of children in a household only offers a protective effect against poverty when using the SPWQ indicator, but according to the MIQ and IEQ indicators you are more likely to be poor. Living in a household with a greater number of employed persons offers protective effect against poverty according to all the three subjective indicators, similarly with having access to electricity. Living in a dwelling that uses flush toilet offers protective effect against poverty according to the SPWQ and MIQ indicators, according to the IEQ

82

indicator you are more likely to be poor. Using piped water in a dwelling and having the refuse removed by a local municipality is associated with less risk of poverty only according to the SPWQ indicator, but according to the MIQ and IEQ indicators you are more likely to be poor. Living in urban informal areas you are more likely to be poor according to the SPWQ and MIQ indicators, however, according to the IEQ indicator your less likely to be poor. Residing in traditional areas is associated with the risk of poverty according to the MIQ and IEQ indicators only, while according to the SPWQ indicator residing in traditional areas offers a protective effect against poverty. Living in rural formal areas only offers a protective effect against poverty according to the MIQ indicator, however, in terms of the SPWQ and IEQ indicators you are more likely to be poor. Residing in a household where the household head is married or cohabiting is associated with less risk of poverty according to the SPWQ and IEQ indicators, but according to the MIQ indicators you are more likely to be poor. Living in a dwelling where there is a fence or wall around the dwelling offers a protective effect against poverty according to the SPWQ and MIQ indicators, but in terms of the IEQ indicator you are more likely to be poor. Living in a dwelling that has a garden is associated with less risk of poverty according to all the three subjective indicators. Being injured in the past month and having required medical treatment is associated with the risk of poverty according to all the three subjective indicators, however, being a victim of crime is only associated with the risk of poverty according to the SPWQ indicator. A higher proportion of males in a household is associated with the higher risk of poverty according to all the three subjective indicators; whereas living in a formal dwelling offers a protective effect against poverty according to all the three subjective indicators. Living in a dwelling that is owned by a household member offers a protective effect against poverty according to the SPWQ and MIQ indicators only, but according to the IEQ indicator you are more likely to be poor.

Annexure A

Annexure A

85

Table: A.1 Logit regressions of South Africans by poverty status and subjective poverty indicator; 2015

Variables

SPWQ MIQ IEQ

Coefficient Standard error Coefficient Standard error Coefficient Standard error Coloured -0,761* 0,002 -0,335* 0,001 -0,496* 0,001

Indian/Asian -0,869* 0,003 -0,292* 0,002 -0,308* 0,002

White -1,290* 0,002 -0,162* 0,001 -0,796* 0,001

Eastern Cape 0,303* 0,002 0,392* 0,001 0,513* 0,001

Northern Cape 0,169* 0,002 0,642* 0,002 0,557* 0,002

Free State 0,054* 0,002 0,381* 0,002 0,221* 0,002

KwaZulu-Natal -0,619* 0,002 -1,534* 0,001 -1,165* 0,001

North West 0,055* 0,002 -0,277* 0,002 -0,133* 0,002

Gauteng -0,098* 0,001 -0,439* 0,001 -0,501* 0,001

Mpumalanga -0,139* 0,002 -0,236* 0,002 0,164* 0,002

Limpopo 0,131* 0,002 0,150* 0,002 0,137* 0,002

Age 0,003* 0,000 0,006* 0,000 0,003* 0,000

Age_Squared -0,000* 0,000 -0,000* 0,000 -0,000* 0,000

Hhsize 0,099* 0,000 -0,035* 0,000 0,005* 0,000

#Children -0,073* 0,000 0,037* 0,000 0,032* 0,000

#Employed -0,290* 0,000 -0,005* 0,000 -0,132* 0,000

Flush_toilet -0,201* 0,001 -0,197* 0,001 0,059* 0,001

Piped_water -0,202* 0,001 0,182* 0,001 0,136* 0,001

Access_electricity -0,267* 0,001 -0,133* 0,001 -0,049* 0,001

Refuse_local -0,051* 0,001 0,011* 0,001 0,167* 0,001

Urban informal 0,222* 0,001 0,152* 0,001 -0,139* 0,001

Traditional area -0,229* 0,001 0,044 0,001 0,168 0,001

Rural formal 0,121* 0,002 -0,058* 0,002 0,069* 0,002

Married_cohabit_head -0,496* 0,001 0,018* 0,001 -0,053* 0,001

Fence_wall -0,483* 0,001 -0,206* 0,001 0,002** 0,001

Garden -0,177* 0,001 -0,168* 0,001 -0,019* 0,001

Injured_treatment 0,178* 0,004 0,017* 0,003 0,053* 0,003

Victim_crime 0,005* 0,001 -0,054* 0,001 -0,016* 0,001

Prop_Males 0,283* 0,001 0,297* 0,001 0,117* 0,001

Formal_dwelling -0,503* 0,001 -0,529* 0,001 -0,257* 0,001

Dwelling_owned -0,186* 0,001 -0,244* 0,001 0,047* 0,001

N 88 647 88 906 86 598

86

Notes:

The data are weighted. Standard errors in parentheses. *P<0.01, **P<0.05. Reference variables: Population group: black African; Settlement type: Urban formal; Province: Western Cape. Hhsize = household size; #Children = number of children below age of 16 in the household; #Employed = number of employed persons in the household; Flush_toilet = dummy variable denoting whether household uses flush toilet; Piped_water = dummy variable denoting whether household source of water is piped water in the dwelling or onsite; Access_electricity = dummy variable denoting whether household has access to electricity; Refuse_local = dummy variable denoting whether household refuse is removed by local municipality; Married_cohabit_head = dummy variable denoting whether household head is married or cohabiting; Fence_wall = dummy variable denoting whether household has a fence or wall around their property; Injured_treatment = dummy variable denoting whether the individual was injured in the past month; Garden = dummy variable denoting whether household has a garden; Victim_crime = dummy variable denoting whether any household member has been a victim of crime in the past year; Prop_ males = proportion of household members who are male; Formal_Dwelling = dummy variable denoting whether the main dwelling is a formal dwelling; Dwelling_owned = dummy variable denoting whether the main dwelling is owned by the household. The models also include a set of variables measuring the level of education attained by the household head not shown in the table.

87

References

Jansen, A., Moses, M., Mujuta, S. C., & Yu, D. (2013). Multifaceted poverty: Absolute, relative and subjective poverty in South Africa. Paper presented at the Economic Society of South Africa, Biennial Conference University of the Free State, Bloemfontein, South Africa. Posel, D., & Rogan, M. (2014). Measured as poor versus feeling poor: Comparing money-metric and subjective poverty rates in South Africa. WIDER Working Paper 2014/133. Helsinki: World Institute for Development Economics Research - United Nations University. Statistics South Africa (2012). Subjective Poverty in South Africa: Findings of the Living Conditions Survey 2008/2009. Report-03-10-01. Pretoria: Statistics South Africa.

Documents

STATS SA 03-10-01... · Findings from the Living Conditions Surveys, 2008/2009 - 2014/2015 Report 03-10-01. Koch Street, Salvokop, Pretoria, 0002 ... is not sufficient to capture